You are on page 1of 185

European Journal of Scientific Research

ISSN: 1450-216X
Editor-In-chief or e Adrian M. Steinberg, Wissenschaftlicher Forscher

Volume 17, No 4 August, 2007

Editorial Advisory Board

Parag Garhyan, Auburn University Morteza Shahbazi, Edinburgh University Raj Rajagopalan, National University of Singapore Sang-Eon Park, Inha University Said Elnashaie, Auburn University Subrata Chowdhury, University of Rhode Island Ghasem-Ali Omrani, Tehran University of Medical Sciences Ajay K. Ray, National University of Singapore Mutwakil Nafi, China University of Geosciences Felix Ayadi, Texas Southern University

Teresa Smith, University of South Carolina Ranjit Biswas, Philadelphia University Chiaku Chukwuogor-Ndu, Eastern Connecticut State University John Mylonakis, Hellenic Open University (Tutor) M. Femi Ayadi, University of Houston-Clear Lake Emmanuel Anoruo, Coppin State University H. Young Baek, Nova Southeastern University Dimitrios Mavridis, Technological Educational Institure of West Macedonia Mohand-Said Oukil, Kind Fhad University of Petroleum & Minerals Jean-Luc Grosso, University of South Carolina

Bansi Sawhney, University of Baltimore Richard Omotoye, Virginia State University David Wang, Hsuan Chuang University Mahdi Hadi, Kuwait University Cornelis A. Los, Kazakh-British Technical University Jatin Pancholi, Middlesex University Jerry Kolo, Florida Atlantic University Leo V. Ryan, DePaul University

As of 2005, European Journal of Scientific Research is indexed in ULRICH, DOAJ and CABELL academic listings.

European Journal of Scientific Research


http://www.eurojournals.com/ejsr.htm Editorial Policies: 1) European Journal of Scientific Research is an international official journal publishing high quality research papers, reviews, and short communications in the fields of biology, chemistry, physics, environmental sciences, mathematics, geology, engineering, computer science, social sciences, medicine, industrial, and all other applied and theoretical sciences. The journal welcomes submission of articles through ejsr@eurojournals.com. 2) The journal realizes the meaning of fast publication to researchers, particularly to those working in competitive & dynamic fields. Hence, it offers an exceptionally fast publication schedule including prompt peer-review by the experts in the field and immediate publication upon acceptance. It is the major editorial policy to review the submitted articles as fast as possible and promptly include them in the forthcoming issues should they pass the evaluation process. 3) All research and reviews published in the journal have been fully peer-reviewed by two, and in some cases, three internal or external reviewers. Unless they are out of scope for the journal, or are of an unacceptably low standard of presentation, submitted articles will be sent to peer reviewers. They will generally be reviewed by two experts with the aim of reaching a first decision within a three day period. Reviewers have to sign their reports and are asked to declare any competing interests. Any suggested external peer reviewers should not have published with any of the authors of the manuscript within the past five years and should not be members of the same research institution. Suggested reviewers will be considered alongside potential reviewers identified by their publication record or recommended by Editorial Board members. Reviewers are asked whether the manuscript is scientifically sound and coherent, how interesting it is and whether the quality of the writing is acceptable. Where possible, the final decision is made on the basis that the peer reviewers are in accordance with one another, or that at least there is no strong dissenting view. 4) In cases where there is strong disagreement either among peer reviewers or between the authors and peer reviewers, advice is sought from a member of the journal's Editorial Board. The journal allows a maximum of two revisions of any manuscripts. The ultimate responsibility for any decision lies with the Editor-in-Chief. Reviewers are also asked to indicate which articles they consider to be especially interesting or significant. These articles may be given greater prominence and greater external publicity. 5) Any manuscript submitted to the journals must not already have been published in another journal or be under consideration by any other journal, although it may have been deposited on a preprint server. Manuscripts that are derived from papers presented at conferences can be submitted even if they have been published as part of the conference proceedings in a peer reviewed journal. Authors are required to ensure that no material submitted as part of a manuscript infringes existing copyrights, or the rights of a third party. Contributing authors retain copyright to their work. 6) Submission of a manuscript to EuroJournals, Inc. implies that all authors have read and agreed to its content, and that any experimental research that is reported in the manuscript has been performed with the approval of an appropriate ethics committee. Research carried out on humans must be in compliance with the Helsinki Declaration, and any experimental research on animals should follow internationally recognized guidelines. A statement to this effect must appear in the Methods section of the manuscript, including the name of the body which gave approval, with a reference number where

appropriate. Manuscripts may be rejected if the editorial office considers that the research has not been carried out within an ethical framework, e.g. if the severity of the experimental procedure is not justified by the value of the knowledge gained. Generic drug names should generally be used where appropriate. When proprietary brands are used in research, include the brand names in parentheses in the Methods section. 7) Manuscripts must be submitted by one of the authors of the manuscript, and should not be submitted by anyone on their behalf. The submitting author takes responsibility for the article during submission and peer review. To facilitate rapid publication and to minimize administrative costs, the journal accepts only online submissions through ejsr@eurojournals.com. E-mails should clearly state the name of the article as well as full names and e-mail addresses of all the contributing authors. 8) The journal makes all published original research immediately accessible through www.EuroJournals.com without subscription charges or registration barriers. European Journal of Scientific Research indexed in ULRICH, DOAJ and CABELL academic listings. Through its open access policy, the Journal is committed permanently to maintaining this policy. All research published in the. Journal is fully peer reviewed. This process is streamlined thanks to a user-friendly, web-based system for submission and for referees to view manuscripts and return their reviews. The journal does not have page charges, color figure charges or submission fees. However, there is an article-processing and publication charge. Further information is available at: http://www.eurojournals.com/ejsr.htm EuroJournals Publishing, Inc. 2005

European Journal of Scientific Research


Volume 17, No 4 August 2007 Contents
Survey of Soil Microarthropods in Selected Sites in Ogun State , Nigeria 433-439

Oke, O.A, C.O.Adejuyigbe and O.A, Ajede


The Toxicity of Hexanolic Extract of Xylopia Aethiopica to Laboratory Reared Larvae of Cx. P. Quiquefasciatus 440-444

Amusan, A.A.S and Oke, O.A


Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan 445-454

Zaheer Abbas and M. Faisal Rizwan


Static Analysis of Slip Damping With Clamped Laminated Beams 455-476

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran


Ontology-Based Framework for Integration in Ubiquitous Computing Environment 476-491

Eduard Babkin and Habib Abdulrab


Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams 491-508

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor 509-522

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee


Intrusions Detection Mechanism by Resilient Back Propagation (RPROP) 523-530

Iftikhar Ahmad, Sami Ullah Swati and Sajjad Mohsin


Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton 531-545

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman
The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector 546-559

Filiz Giray
Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE 560-573

aban elik
Ethnic Violence in Sindh and its Repercussions on South Asia 574-584

Amir Ali Chandio and Ishtiaq Ahmed Choudhry


Influence of Supervisors Gender on Mathematics Student Teachers Performance in Teaching Practice 585-591

Ugboduma, S. O and EZE, A.E

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) : Exemple des Commune DAbobo, Adjam, Attcoub, Cocody et Port-Bout

592-604

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne


Double Acceptance Sampling Based on Truncated Life Tests in Rayleigh Distribution 605-611

Muhammad Aslam

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.433-439 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Survey of Soil Microarthropods in Selected Sites in Ogun State, Nigeria


Oke, O.A Department of Biological Sciences University of Agriculture Abeokuta C.O.Adejuyigbe Department of Soil Sciences and Agricultural Mechanization University of Agriculture, Abeokuta O.A, Ajede Department of Biological Sciences, University of Agriculture Abeokuta Abstract A survey of the soil fauna of selected farm locations in Ogun State was conducted. The study was carried out at the school farm of the University of Agriculture, Alabata, Abeokuta; the School garden of the Baptist Girls` College, Idi-Aba, Abeokuta; a Farmland at Ewekoro, the Rubber plantation, Ikenne. All sites are in Ogun State. Microarthropods encountered during the survey of the sites considered belong to Class Arachnida, order Acarina, Suborders Oribatida, x = 36.50; Actinedida x = 1.00 and Gamasida; x = 2.25 and class insecta, order Collembola, x = 0.75. The rubber plantation was found to have the highest population of all these Microarthropods x = 24.75 while the school farm of the University of Agriculture, Abeokuta recorded the least population containing species of Suborder Oribatida x = 1.75 only. The school garden of Baptist Girls` College, Idi-Aba, Abeokuta x = 11.75 and the Farmland at the Cement Factory area, Ewekoro x = 3.75. The result of the physico-chemical analysis of the soil conducted indicated the influence of soil properties on the distribution of these microarthropods, also, different farm practices.

Introduction
Soil fauna communities include soil inhabiting invertebrates. They are known to improve soil structure by decreasing bulk density, increasing soil pore space, soil horizon mixing, increase aeration and drainage, increase water holding capacity, liter decomposition and soil aggregate structure (Abbott, 1989). The soil organisms range from macroscopic forms to the microscopic forms (Franke, 2003). There are two size classes of invertebrates that live in the soil. They include microfauna such as protozoa, mesofauna (which include nematodes, microarthropods and diptera larvae), and macro fauna, which include macro arthropods and earthworm (Abbott,1989). Population density and composition of the fauna in soils are indicators of soil condition (Lussonhop, 1992); (Stork and Eggleton, 1992) and rehabilitation of ecosystem quality because they can integrate information about other values as well as soil quality (Curry, 1979). Adejuyigbe et al., (1999), observed that soil degradation due to physical disturbance associated with cultivation, depletion of soil organic matter, reduced floral diversity and absence of plant cover for part of the year leads to depletion in the population of soil microarthropods.

Survey of Soil Microarthropods in Selected Sites in Ogun State , Nigeria

434

For an area with degraded soil such as mined or intensely cultivated sites, the re-establishment of invertebrates can simply be a matter of removing the cause of degradation or may involve more extensive site preparation techniques and re- introduction (Bonnie, 2003). Killham (1995), stated that the abundance and diversity of soil animals is greatest in soil of little or no disturbance such as those of permanent grassland and woodland, where they play a vital role in nutrient cycling. Tropical savannahs and forest have a lower fauna biomass than their temperate counterpart David et al., (1989). It further explained that for forest ecosystems, the fauna biomass is often inversely related to the accumulated weight of organic matter in the soil surface. Climate also influence the composition of soil fauna. Brussaard et al. (1997) observed that physical and chemical characteristics of the soil are prominent soil are prominent soil characteristics determining the occurrence and functioning of the microarthropods. The abundance of microarthropods is determined by resource availability (i.e. organic matter), macro and micro climate, pH and disturbance. (Petersen, 1982; Curry, 1994). The objectives of this study are therefore to survey the soil fauna microarthropods in different soil type and land use and to assess the influence of both physical and chemical factors of the soil on microarthropod population and distribution.

Materials and Methods


The sampling sites where soil samples were collected were School farm, University of Agriculture Alabata Abeokuta; School garden of Baptist Girls` College, Idi-Aba, Abeokuta; Farmland at Cement factory area, Ewekoro and Rubber plantation farm settlement, Ikenne, all in Ogun State, Nigeria. Samples of each soil were collected to 10.cm depth with sampling core. Growing vegetation was gently removed using cutlass, root and leaf debris were also removed. The samples were air dried for laboratory analysis. Similar core samples were taken at each sites for microarthropods extraction. Microarthropods in the sample were collected using core of size 50cm2 from 0-10cm soil depth. Without removing the liter layer, microarthropods were extracted using a modified Tullgreen type extractor with a 25-watt bulb suspended over each sample for three days. Soil microarthropods were collected in formalin solution, their abundance were sorted and counted using a stereomicroscope and were grouped into taxonomic groups. Photographs of microathropods were taken using photomicrogramm. Particle size analysis introduced the dispersion and segmentation rate and the use of hydrometer. The sand fractions were obtained by deducting the values obtained for silt and clay fractions from 100%. The classes of the soil were traced out on a textural class triangle. Organic matter was derived from organic carbon using Walkey Black method. Soil pH was determined using digital pH meter. Fifty grammes portion of fresh soil samples were weighed from each samples into a can of known weight. The soil samples in the cans were oven dried at 105oC for 24 hours. The final weight of the cans and the content were weighed after drying. Moisture content is given by = Moisture loss after heating X 100 Weight of dry soil Temperature of each sample taken was determined by inserting a mercury glass bulb thermometer into the tied polythene bag containing the soil samples. Temperature readings were taken at a point at which the reading is stable. Temperature readings of the four samples taken from each sites surveyed were recorded. ANOVA, Multiple Regression and Pearson Coefficient Correlation were used for the Statistical Analysis.

Results
Population of Microarthropods in the Different Soil Samples Microarthropods population found were of class Arachnida, order Acarina, suborders Oribatida (Cryptostigmata); Microarthropods population was found to be highest in rubber plantation and lowest

435

Oke, O.A, C.O.Adejuyigbe and O.A, Ajede

in UNAAB farm. For the entire soil sample, the microarthropods population was not significantly different (P > 0.05) from each other with the exception of gamasids population. Soil of rubber plantation was observed to have highest species biodiversity or richness while the University of Agriculture Abeokuta farm was found to be least in species abundance. Table 1 shows the mean population of microarthropods in the different soils.
Table 1: Mean population of microarthropods taxonomic taxa in the different soil samples.
Class arachinda order acarina Sub Orders Oribatida Actinedida Gamasida 1.75a 0.00a 0.00b 11.00a 0.75a 0.00b 3.25a 0.00a 0.50b 20.50a 0.25a 1.75a 36.50 1.00 2.25 Class Insecta Order Collembola 0.00a 0.00a 0.00a 0.75a 0.75 Total 1.75a 11.75a 3.75a 24.75a 42.00

Sampling Site Unaab Farm School Garden Idi-Aba Farm Land at Ewekoro Rubber Plantation

Note: Values in each column with the same letter are not significantly different (P > 0.05).

Particle Size Distribution School garden of Baptist Girls` College, Idi-Aba has the highest percentage of 92.79% sand while farmland at Ewekoro has the lowest value of sand (80.20%). The percentage of clay and silt in the soil were generally low. The percentage of clay was found to be high in soil with low percentage of sand and low in soil with high percent.sand. The highest percentage of clay was observed in soil from farm land at Ewekoro. The soils fall into two textural classes that is sand and loamy sand. Table 2 shows the particle size distribution.
Table 2: Particle size distribution
Clay% 4.95a 3.63a 11.48a 10.45a Silt% 5.05a 3.57a 8.32a 3.87a Sand% 89.97a 92.79a 80.20a 85.65a Tectural Class Sand. Sand. Loamy sand Loamy sand

Site Unaab Farm School Garden Idi-Aba Farmland at Ewekoro Rubber Plantation
Note:

Values in each column with the same letter are not significantly different (P > 0.05).

Distribution of Soil Organic Matter The organic matter at the School garden in Idi-Aba was observed to be the least. It was observed that organic matter of the University of Agriculture Abeokuta farm and the Rubber plantation at Ikenne are not significantly different (P > 0.05) from each other but significantly different (P < 0.05) from organic matter of the School garden at Idi-Aba and the Farmland at Ewekoro. Organic matter of the School garden of Baptist Girls College Idi-Aba and the Farmland at Ewekoro are significantly different (P < 0.05) from each other. The result in Table 3 showed that organic matter of the Farmland at Ewekoro was the highest followed by the Rubber plantation at Ikenne.
Table 3: Distribution of soil organic matter
Organic Matter (%) Sample 2 3 2.29 2.41 1.81 2.59 3.86 3.66 1.86 2.83

Site Unaab Farm School Garden Idi-Aba Farmland at Ewekoro Rubber Plantation

1 2.26 1.31 2.55 4.83

4 1.24 2.02 3.04 3.10

Mean O. M 2.18a 1.93b 3.27a 3.15ab

Note: values in each column with the same letter are not significantly different (P > 0.05).

Survey of Soil Microarthropods in Selected Sites in Ogun State , Nigeria pH of The Soil Considered

436

It was observed from the result that the University of Agriculture, Abeokuta farm, the School garden of Baptist Girls College, Idi-Aba and the Rubber plantation at Ikenne soils were acid soils but pH of University of Agriculture, Abeokuta farm and the School garden Idi-Aba is significantly different (P < 0.05) from pH of the Rubber plantation Ikenne. Soil of farmland at Ewekoro was observed to be alkaline (Table 4). Table 4: pH of the different soil
Site Unaab Farm School Garden at Idi-Aba Ewekoro Rubber Plantation 1 5.92 7.81 6.16 5.36 2 6.40 5.86 7.20 6.97 3 5.87 5.88 7.34 5.17 4 5.95 7.36 9.13 5.70 Mean Soil pH 6.04ab 6.73ab 7.46ab 5.80b

Note: values in each column with the same letter are not significantly different (P > 0.05).

Moisture Content of The Soil Considered The moisture content of the University of Agriculture, Abeokuta farm soil was extremely low while that of the Farmland at Ewekoro and the Rubber plantation, Ikenne were the highest. With the School garden Idi-Aba however, the moisture content was found to be lower (Table 5).
Table 5: Moisture content of soil considered
1 0.89 1.21 4.36 14.81 % Moisture Content 2 3 0.83 0.81 1.0 1.42 5.62 14.97 2.31 7.33 4 0.60 2.04 4.04 7.25 Mean Moisture Content 0.78b 1.42b 7.23a 7.90a

Sampling Site Unaab Farm School Garden at IDI-ABA Farmland AT Ewekoro Rubber Plantation

Note: values in each column with the same letter are not significantly different (P > 0.05).

Temperature of the Soil Considered It was observed from the result that the University of Agriculture, Abeokuta farm has the highest temperature while temperature of rubber plantation was the lowest. The range temperature between farmland at Ewekoro and school garden is 10c (Table 6).
Table 6: Temperature of the soil considered
1 33.8 33.20 32 26 2 34.3 33.50 31 26 3 34.0 33.50 31 30 4 34 33 35 29 Mean Temperature (C) 34.05a 33.43a 32.25a 27.75b

Sampling site Unaab Farm School Garden at IDI-ABA Farmland at Ewekoro Rubber Plantation

Note: Values in each column with the same letter are not significantly different (P > 0.05).

Correlation Between Particle Size, Moisture Content, Organic Matter, Temperature and Soil pH Organic matter correlated negatively with temperature, this is significant at 5%, so also temperature correlated negatively with moisture, which is significant at 5%. Table 7 showed how the parameters correlated with each other either negatively or positively which can also be significant or nonsignificant at 1% and 5% significant levels.

437
Table 7:
Clay Silt Sand Moisture

Oke, O.A, C.O.Adejuyigbe and O.A, Ajede


Correlation between particle size, moisture content, organic matter, temperature and soil pH
Organic matter 0.479 0.650** -0.596* 0.8135 Temperature -0.338 -0.071 0.274 -0.614* pH 0.109 0.124 -0.129 -0.075 Clay 1.000 0.559* -0.951** 0.785** Silt 0.559* 1.000 -0.788** 0.569* Sand -0.951** -0.787** 1.000 -0.792**

Note: *5% significant level **1% significance level.

Correlation Between Microarthropod Populations, Particle Size, Temperature, Moisture Content, pH and Organic Matter The total microarthropods population correlates significantly with moisture. Oribatids population correlates negatively with temperature, this was significant at 1%. It also correlates significantly with moisture. Table 8 shows the coefficient correlation between microarhtropod populations, physical and chemical properties of the soil considered.
Table 8: Correlation between microarthropod populations, particle size, temperature, moisture, pH and organic matter
Clay 0.155 0.186 Slit 0.066 0.073 Sand -0.134 -0.159 Temp. -0.626** -0.654** Organic matter 0.486 0.511* Moisture 0.529 0.564* pH -0.168 -0.198 Oribatid 1.000** 0.995** Total 0.996** 1.000

Oribatids Total

Note: *5% significant level **1% significance level.

Dependent Variables (Physical and Chemical Proprtties Of Soil) For Microarthropods Population For total microarthropods population, all the variables are non-significant except temperature which is significant at 1%. Using multiple regression analysis, all the variables are non-significant together in functioning of total microarthropods population. Through backward elimination however pH was the first variable eliminated yet the remaining variables are non significant. pH elimination was followed by organic matter removal with the remaining variables close to significant level. Elimination of moisture made temperature only to be significant in functioning of microarthropods. Hence order of dependence for total microarthropods is temperature moisture organic matter pH. This is also the case for Oribatids population with slight change in order of dependence of variables. The order of dependence for Oribatids is temp -- organic matter-- pH. Table 9 is a result of multiple regression analysis for total microarthropods population showing order of variable dependence of microarthropods.
Table 9: Dependent Variables (Soil Properties) For Microarthropods Population
Oribatids 0.931 0.374 0.777 0.009** Total 0.832 0.712 0.327 0.006**

Dependent variable pH Organic matter Moisture Temperature


Note: *5% significant level **1% significance level.

Discussion
This work was carried out with the aim of surveying the varying land farm practices sites chosen for their microarthropod distribution and the influence of the soil properties considered on the microarthropod abundance. This is attributed to the land use and the physicochemical properties of the

Survey of Soil Microarthropods in Selected Sites in Ogun State , Nigeria

438

soil supporting the evidence that environment, physical and chemical properties of the soil affect the distribution and abundance of invertebrates (microarthropods) directly by influencing their biology and life processes and indirectly by determining the nature of the habitats they occupy and their food supply (Curry, 1994). It was found that the soil properties considered differed from one site to another. The result showed that the variation in sand, silt and clay content was not very high and the soil of the four sites considered fall into the textural classes i.e. loamy sand soil and sand soil. The Rubber plantation had high organic matter. This could be attributed to its continuous supply of organic material for decomposition. The UNAAB farm organic matter is statistically similar to the rubber plantations organic matter due to its fertilizer inputs. Considering the soil pH, soil of the farm land at Ewekoro was slightly alkaline due to the effect of calcium from lime stone found in the dust of the environment. Other sites were acidic soil though the rubber plantation soil was statistically different from the rest. Temperature of the soil range from 27.75 -- 34.05 with the rubber plantation having the least because rubber trees produced a cover (shade) to the soil while the UNAAB farm had the highest temperature value because the soil was exposed to direct sunlight (crop had been harvested). It was observed that the presence of rubber trees and arable crops such as sugarcane and cassava in rubber plantation and the farm land at Ewekoro respectively shield the sun from direct penetration or sunlight hence high moisture content while the reverse was the case for the UNAAB farm and the school garden of the Baptist Girls College, Idi-Aba. Results also showed that properties considered correlates significantly with each other with few exceptions. This is in substantial agreement with those of Curry (1994) that all the physical and chemical properties of soil operate strongly interconnected. Soil microarthropods bear greater relationship with temperature. As temperature increases, soil microarthropod population decreases. This is due to the fact that temperature directly affect the rate of physiological reactions and indirectly has effect on the soil biological activities, also soil animals migrate to avoid high temperature. This observation supports the view of Swift et al., (1979) that soil temperature is a factor of paramount importance in terms of distribution and activities of the soil animals. The population of total microarthropods was found to increase as organic matter increase. Moisture also correlates significantly with soil microarthropods mainly because it influences soil water content and relative humidity of the soil pores. Result has shown that moisture correlated significantly with organic matter. Both moisture content and organic matter correlated significantly with soil microarthropods. This signifies that apart from microarthropods feeding on organic matter, it also aids microbial growth providing food for the microarthropods, hence influencing microarthropod population. This observation positively support the view that moisture increases the surface area of organic matter particles available for microbial attack thus increasing the growth rate of bacteria and fungi use as food by microarthropods (Visser, 1985). The rubber plantation has the highest population density of soil fauna and species biodiversity. This could be attributed to land use. This supports the view that perennial crops offer the most suitable conditions to microarthropods compared with arable fields. In the rubber plantation, the soil micro flora, the main primary food source of the microarthropods is stimulated by continuous inputs of organic matter from the liter, roots and often manure all year round. Also the absence of soil tillage constitutes the main difference between annual and perennial crops. Favorable soil characteristics also influence the high population density. The School garden soil, which was less degraded, had high population density of soil fauna with its major population belonging to suborder Oribatida. Although the soil is acidic with low organic matter, high temperature and low moisture content the soil still support soil fauna because it was less disturbed. Despite the favorable soil characteristics of soil characteristics of farm land at Ewekoro, the soil could not support large number of soil fauna as found in the school garden at Idi-Aba because it is subjected to continuous cultivation. This is in agreement with those of Adejuyigbe et al., (1999) that soil degradation due to physical disturbance associated with cultivation, reduced flora diversity. Also, absence of plant cover for a part of the year leads to depletion in the population of soil microarthropods. Its high organic matter was signified by the number of soil millipedes obtained.

439

Oke, O.A, C.O.Adejuyigbe and O.A, Ajede

Oribatid mites or beetles mites was the only soil fauna observed in the soil from the UNAAB farm. This could be attributed to the fact that crop on the land had been harvested, exposing the soil to direct effect of sunlight. Oribatid mites were found to be common to the four sites. For the total microarthropods population, results from this study has shown that soil temperature contributed significantly to its abundance, organic matters mainly because some of the microarthropods live and feed in liter layer. Influence of moisture content on the total Oribatids abundance was not significant while effect of pH contribution was the least and of no significance.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] Abbott, I. (1989). The influence of fauna on soil structure. In majer, J.D. (ed) Animals in Primary Succession: the role of fauna in reclaimed lands. Cambridge University Press. Pp 3950 Adejuyigbe, C.O., Tian, G. And Adeoye, G.O. (1999). Soil microarthropods populations under natural and planted flowers in Southwestern Nigeria. Agroforestry system 47: 263-272. Bonnie, W. (2003). Using Soil Fauna to improve Soil Healths. //www.nhm.ac.uk/science/entom/project3. Brussaard, L., Hassink, J. and Vreeken-Buijs, M. J. (1997). Relationship of Soil Microarthropod Biomass with Organic matter and pore size Distribution in soils under Different Land Use. Soil Biol. Biochem. 30. No 1. Pp 97. Curry, J.P. (1979). The Arthropods Fauna Associated with Cattle Manure Applied as Slurry to Grassland. Procedings of the Royal Irish Academy, 76B, 49-71. Curry, J.P. (1994). Grassland Invertebrates. Chapman and Hall. London. David, C.C., Malcolm, J.O. and Gorouehara, B. (1989). Dynamics of Soil Organic Matter in Tropical Ecosystem. Mc Graw Hill. London. Pp 1-19, 62-64. Franke, B. (2003). Arthropod fauna of soils. News letter of the Biological survey of Canada (Terrestrial Arthropods) 20, No.2. Killham, K. (1995). Soil Ecology. Cambridge. University press. London. Lussenhop, J. (1992). Mechanisms of microarthropod-microbial interactions in soil. Adv Ecol. Res.23, 1-33. Peterson, H. (1982). The soil fauna biomass and its composition. Oikos 39: 330-339. Robert, L. T. (1995). Soil Microbilogy. John Willey and Sons Inc. USA. Pg 38-42. Seastedt, T.R. (1984). The Role of Microarthropods in Decomposition and Mineralization Processes, Ann. Rew. Entomol. 29: 25-46. Stork, N.E. and Eggleton, P. (1992). Invertebrates as determinants and indicators of soil quality. American Journal of Alternative Agriculture. 7; 38-45. Swift, M. J., Heal, O.W. and Anderson, J.M. (1979). Decomposition in Terrestrial Ecosystems. Blackwell. London. Visser, S. (1985). Role of soil invertebrates in determining the composition of soil microbial communities. In: Fitter A.H., and Atkinson, O. (edS). Biological interaction in soil. Blackwell, Oxford. 297-317.

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.440-444 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

The Toxicity of Hexanolic Extract of Xylopia Aethiopica to Laboratory Reared Larvae of CX. P. Quiquefasciatus
Amusan, A.A.S Department of Biological Sciences, College of Natural Sciences University of Agriculture, Abeokuta, Nigeria Oke, O.A Department of Biological Sciences, College of Natural Sciences University of Agriculture, Abeokuta, Nigeria Abstract The hexanolic extract of Xylopia aethiopica (Ethiopian pepper) was tested for acute toxicity on the larva of Culex p. quiquefasciatus reared in the laboratory of the Department of Biological Sciences, University of Agriculture Abeokuta, Nigeria. Five concentrations: (50ppm, 100ppm, 200ppm, 300ppm, 400ppm) were evaluated for acute toxicity on the larvae and total percentage mortalities recorded at intervals of 1, 12, 24, 48 and 96 hours in each test. Effects of sunlight and ultra violet radiation on stability of the extracts potency at 2, 4 and 8 hours respectively were equally examined. The mean lethal concentration LC50 was 257ppm.. Toxicity of Xylopia aethiopica on Culex p. quiquefasciatus larvae was gradual and persisted throughout the test period. Sunlight exposure has no significant effect on the potency of Xylopia oil while ultra-violet radiation activated the larvicidal properties. Keywords: Xylopia aethiopica, acute-toxicity, larvae, Cx. P. quiquefasciatus, lethal concentration, LC50 and potency.

Introduction
Insects have attacked plants since the Devonian period. During this prolong interaction, many plant taxa evolved higher sophisticated defence system, largely a complex array defence chemical produced by the plants themselves. (Hedin, 1991). Renewed interest in botanical pesticides is motivated by three major objectives: (i) To encourage traditional use of simple formulation of locally available plant materials in pest control. To identify sources of new botanical pesticides for commercial extraction. (ii) To elucidate the chemical structure of active components of botanicals. This was based on the fact that novel metabolism may serve as a model for the chemical synthesis of new pesticides with more desirable properties, for example, the known structure of natural pyrethroids have provided a basis for the development of a range of insecticides of great practical value (Elliott et al., 1978). However, in spite of the fact that a variety of secondary metabolites of plant origin afford protection against insects, only a few of these have been developed into commercial insectides. These include margosan-0, developed from Neem, (Larson, 1987). Black leaf 40" produced from the byproduct of tobacco industry (Merck index, 1989), Ryanex or Ryaniacide obtained from Ryania speciosa. Other bioactive compounds of plant origin that has been in use in pest control include

The Toxicity of Hexanolic Extract of Xylopia Aethiopica to Laboratory Reared Larvae of Cx. P. Quiquefasciatus

441

rotenone from Derris guinolizidine, alkaloid from many leguminosae such as Lupinus, Quassin, and Pyrethrum. It is likely that systematic survey of plant materials with the broad range of bioassay using different test species would lead to the discovery of new active compounds although one may not always come across compounds possessing toxicity against insect pests comparable to the very potent currently available insecticides. Hence there is need for continuous search for new compounds with desirable properties. Studies on the dried fruit of Xylopia aethiopica have led to the isolation of a new diterpene acid, Xylopic acid (Ekong and Organ, 1968) and five other kaurane diterpenes (Ekong, 1968). Xylopia essential oil obtained by direct steam distillation of the derived and crushed fruits gave a strong positive reason for aldehydes with Tollens reagent. Oji et al., (1991) reported that laboratory treatment of maize grain with dust or ethanolic extracts of P. guineense and X. aethiopica caused significant mortality of adult Sitophilus zeamais infesting grain. X. aethiopica has not been tested on insects such as mosquitoes. Majority of the work earlier done on Xylopia had been stored products insects, the present study aims at investigating the insecticidal properties of Xylopia aethiopica on larvae of Cx. P. quiquefasciatus in the laboratory.

Materials and Methods


Plant extract Dry fruits of Xylopia aethiopica (Ethiopian pepper) were bought from a local market in Ile-Ogbo, Osun State, Nigeria. The dried fruit were washed and later sun dried. The dried fruit were ground to a fine powder using electric grinder and later exhaustively extracted with n-hexane using Soxhlet apparatus. The extract was then concentrated with rotary evaporator which removed the hexane component leaving behind viscous oil required for the analysis. Preparation of volume/volume stock solution of Xylopia aethiopica was by measuring out 1ml of the extract, emulsify it in about 0.003ml or 3 drops of Tween - 80. The emulsified extract is then added up to 1 litre to form 100ppm stock solution. From the stock serial concentrations of; 50ppm, 100ppm, 200ppm, 300ppm, and 400ppm were prepared. Insect material Sands and debris from breeding sites of Cx. P. quiquefasciatus were collected at the peak of dry season(January) and soaked in water. After a day, hatching of viable eggs commenced. Hatched out larvae were then removed in to cleaner water inside brightly coloured enamel plates using a pipette. Active swimming non- feeding pupae were thereafter collected into open bottles and placed in a cage where they are left to emerge. Mosquito cages of about 40 by 40 cm were made of light wooden frame with sides of mosquito netting and base of wooden board. One side of the cage is provided with sleeve for taking material in and out of the cage. Adult mosquitoes were allowed to feed on rabbits kept in cages after they have been starved for one to two days. Oviposition commenced after three days of the first blood meal and lasted for about seven days. Eggs were collected, dried, and stored away in an incubator at 31+2C. All the third instars larvae used in the bioassay were hatched out from the stored eggs. Evaluation of the Effect of Xylopia Aethiopica Fruit Oil Cx. P. Quiquefasciatus Larvae From each concentration 250ml solution was measured and introduced into separate labeled 500ml specimen bottles. Forty larvae of Cx. P. quiquefasciatus were then introduced into each. The bottles were then covered mesh of 1mm for aeration but prevents entry and exit of any insect. Each treatment was replicated five times. Effects of physicochemical parameters such as sunlight and ultraviolet irradiation were tested on the extract using the method described by Adewunmi and Marquis (1980). A

442

Amusan, A.A.S and Oke, O.A

stock solution of the X. aethiopica was exposed to sunlight and a U.V lamp (Gallenkamp LH530) with peak output at 366nm for 2, 4 and 8 hours. For the U.V light, stock solution was placed at a distance of about 30cm from the light source. Mortalities were recorded at intervals of 1, 12, 24, 48 and 96 hours. To estimate the 96-hour median lethal concentration (LC50) of the extract , the 96 hours mortality of the different concentrations were used from these data using probit regression line. Data obtained from processes described above were then subjected to analysis of variance at 5% level of significance and where there was a difference, Duncans test was used to determine whether there were significant differences between treated and untreated means.

Results
The acute LC50 of Xylopia aethiopica oil extract was 257ppm (Table 1) very few deaths were recorded in the first 6 hours, percentage mortality increased with increase in time. The highest concentration of 400ppm killed only 5% of the population in 1 hour test period, 3% in 6 hours and 75 % in 24 hours test period. Toxicity of X. aethiopica was observed to be gradual and lasted throughout the test period.
Table 1: LC50 at 96 hours of Xylopia aethiopica applied to Cx. P. quiquefasciatus larvae.
Concentration 257ppm 200ppm 199.53ppm 316.23ppm 63.1ppm 95.5ppm 63.1ppm

Treatments Untreated 2 hours sunlight 4 hours sunlight 8 hours sunlight 2 hours U.V 4 hours U.V 8 hours U.V

Duncans test revealed that X. aethiopica oil extract exposed to sunlight showed no significant difference between the untreated X. aethiopica oil extract and 2, 4 and 8 hours sunlight exposure. The comparative LC50 value presented in Table 1 revealed little differences between the treated and untreated extracts except for 8 hours treatment. Within the first 1 hour of treatment, no larvae death was recorded. At 2 hours, 60% mortality was recorded, while 47% and 63% were recorded at 4 and 8 hours exposure respectively, compared to 75% mortality value recorded for the unexposed X. aethiopica oil extract. While considering the effect of ultraviolet irradiation on the potency of X. aethiopica, result reveals that at 1 hour test period, 40%, 60% and 45% mortalities were recorded for 2, 4 and 8 hours exposure period respectively. Compared to 5% mortality recorded for unexposed X. aethiopica oil extract. Duncans test showed significant variation in their means (P>0.05) (Table 2). The treated Xylopia aethiopica oil extract was significantly different from the untreated. Variation of the LC50 showed an activation of the potency of Xylopia aethiopica on irradiation but without a definite trend.

Discussions
Toxicity of Xylopia aethiopica hexane extract was gradual and persisted throughout the test period of 96 hours. The toxicity and persistence of X. aethiopica hexane extract is very reassuring in the control of Cx. P. quiquefasciatus in the field since they lay their eggs on a daily basis when and where breeding exist. The persistence will reduce the need for a daily repeated application. The toxicity of Xylopia aethiopica extract to larvae of Cx. P. quiquefasciatus corroborate the report of Oji et al.(1991) that X. aethiopica showed significant mortality effects on adults Sitophilus zeamais.

The Toxicity of Hexanolic Extract of Xylopia Aethiopica to Laboratory Reared Larvae of Cx. P. Quiquefasciatus

443

Ekong and Ogan (1968) isolated a diterpene acid, Xylopic acid, which could be the major insecticide present in X. aethiopica. However the mode of action of Xylopic acid is still obscure. Sunlight exposure has no significant effect on the potency of X. aethiopica oil extract. This stability was a deviation from Winks (1993) report that pesticide principles of plant origin are unstable under sunlight. Ultraviolet (U.V) portion of the sunlight was observed to activate the larvicidal properties of X. aethiopica. The findings in this study is in line with observations of Graham et al.,(1980), Wat et al., (1979) and Arnason et al., (1981) works where sunlight was found to significantly affect the potency of various botanical pesticides. Since X. aethiopica is very common in the local market, it could be a very cheap source of a new pesticide. However, further research is necessary to identify the active ingredient in Xylopia aethiopica as well as its mode of action.
Table 2:
Sunlight

Comparison of treatment mean (Duncans test)


Untreated 2 hours 4 hours 8 hours LSD Untreated 2 hours 4 hours 8 hours LSD 4.600a 5.250a 4.950a 6.050a 2.289 4.600a 6.600a 4.650b 3.35b 1.7467

U.V

*Means followed by the same letter are not significantly different

444

Amusan, A.A.S and Oke, O.A

References
[1] Adewunmi, C.O, Marquis, V.O and Sofowora, E.A (1980) Preliminary screening of some plant extracts for moluscicidal activity. Plant medica, 39: 57 - 65 [2] Arnason, T., Swain, T., Wat, C.K, Graham, E.A., Partingoto, S. and Towers, G.H.N (1981) Mosquito larvicidal activity of poly acetylenes from species of the Asteraceae. Biochem Systematics Ecol. 9:63-65. [3] Ekong, D.E.U and Ogan, A.U (1968) Chemistry of the constituents of Xylopia aethiopica. The structure of Xylopic acid, a new diterpene acid. J. Chem. Soc. 3:311-312. [4] Elliot, M., Janes, N.F and Pottes, C (1978). The future of pyrethroides in insect control. Ann. Rev. Entomol. 23:443-469. [5] Graham, K., Graham, E.A and Towers, G.H.N (1980). Cercaricidal activity of phenyl heptatriyne and terthienyl naturally occurring compounds in species of Asteraceae (Compositae). Can J. Zool. 58(11)1955-1958. [6] Hedin, P.A (1991) Use of natural product in pest control in naturally occurring pest Bio regulators. A C S Symposium Series No .449:162-171. [7] Larson, R.O (1987) Development of Margosan-O, a pesticide from Neem seed. In proceedings 3rd int. Neem conf., Nairobi. Edited by Schitterer, H. and Ascher, K.R.S. pp 243-250. [8] Oji, O., Madubuike, F.N, Nwaigbo, L.C and Oji, P.O (1991). Protection of stored Zeamays against Sitophilus spp with non-toxic natural products. Potentials of Xylopia aethiopica and Piper guineense. Acta agronomica, Hungaria. 41(1-2)131-135. [9] Wat, C.K, Biswas, R.K, Graham, E.A, Bohm, L and Towers, G.H.N (1979). Ultraviolet mediated cytotoxic activity of phenyl heptatryne from Bidens pilosa. L. J. Nat prod. 42(1):103-111. [10] Wink, M (1993) Production and application of Phytochamicals from an Agricultural perspective. Proc. of phytochemical Soc., Europe 34. Phytochemical and Agriculture (T.A Van Beek and H. Breteher eds).

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.445-454 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan
Zaheer Abbas Mohammad Ali Jinnah University Islamabad M. Faisal Rizwan Lecturer GC University Faisalabad Mohammad Ali Jinnah University Islamabad Abstract Banking industry all over the world is subject to comparatively increasing regulation and, simultaneously, finds itself under the close scrutiny of regulators and other stakeholders. For example in case of Pakistan, under supervision of State Bank of Pakistan and other stake holders including, Securities and Exchange Commission of Pakistan (SECP) and large market analysts. Banks are required to maintain minimal equity relative to both total and risky assets. Market analysts expect banks to grow at a certain rate and to show reasonable returns on assets and on equity. The requirements of capital adequacy ratio on one hand and expectations regarding high profit on the other hand create incentives for management to hide earnings in good time and increase reported profit in bad times with the help of accrual management. Thus, it is of due importance to study the effect of these tactics (accruals) on the value of the bank. Value of Bank has been estimated using Tobins Q formula while total accruals have been measured through income statement approach i.e. difference between net income and net cash flows from operating activities. In this study, Cross Sectional Modified Jones Model [1995] has been used to segregate total accruals into discretionary and nondiscretionary components. Using ordinary least square method (OLS), in a sample of 21 commercial banks operating in Pakistan, I document negative relationship between discretionary accruals and firm value and very strong positive relationship between firm value and profitability of bank in a particular year. However, relationship of firm value with total accruals has been found positive but insignificant. These findings are the indicators that commercial banking firms face negative response if they deliberately report less earnings in good times and comparatively higher earnings in bad times.

Introduction
Banking industry all over the world is subject to comparatively increasing regulation and, simultaneously, finds itself under the close scrutiny of regulators and other stakeholders. For example in case of Pakistan, under supervision of State Bank of Pakistan and other stake holders including, Securities and Exchange Commission of Pakistan (SECP) and large market analysts. Banks are required to maintain minimal equity relative to both total and risky assets, currently 8% according to Basel Capital Accord II. Market analysts expect banks to grow at a certain rate and to show reasonable returns on assets and on equity. The requirements of capital adequacy ratio on one

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan

446

hand and expectations regarding higher profit on the other hand create incentives for management to hide earnings in good time and increase reported profit in bad times. Thus it is of due importance to study the effect of these tactics on firm value. Further, this study determines whether corporate management is successful or not in misguiding the market players through accounting tactics. In summary, this study will help in finding out the contribution of accounting tactics and firm economic variables in determining the value of bank in financial market of Pakistan. A brief view of financial statements of banks raises doubts regarding sudden changes in the presentation of financial statements and some questions get more importance. For example, had the accounting method of computation of provision for irrevocable advances not been changed, profit and advances of Bank Alfalah for year 2004 would have been higher by Rs. 95.657 million. Furthermore, during year 2004, Bank Al Falah changed rate of depreciation on computer equipment from 20%-25%. Had this change not been made, operating fixed assets would have been higher by Rs. 12.239 million. Had the method of revaluation of investments (held for trading) not been changed, investments and surplus on revaluation of investments of Bank Al Habib would have been higher by Rs. 22.874 million and Rs. 43.790 million respectively. Lambert [1984] suggests that devices for earning smoothing can be classified into two categories. One is real actions technique (e.g. derivatives, which affect the actual cash flows) and the second is artificial techniques (e.g. discretionary accruals). Both of these techniques contribute in earning management, one on actual cash flow basis and second on accrual basis. As derivatives income is also included in calculation of net income so its presence in operating activities does not constitute any component of accruals. Furthermore, these actual cash flow based techniques have very little contribution in earning management in Pakistan as their usage is still limited to only foreign exchange markets and very low even in equity markets. Total accruals have been divided into its two components using Cross Sectional Modified Jones Model [1995]. One component is discretionary accruals while second one is non-discretionary component. Both components have been used as independent variables in separate regressions to capture their contribution in determining market value and highlighting management objective in manipulating capital market and its players. Whether management successfully increases the bank value through these accounting tactics or not? The question is very important because it provides insight into managements objective to increase value of their firm through their so-called accounting tactics. Further, it sheds light on whether management is able to increase the value of their firm through these tactics or fail and get negative response. Investors attitude has been tested whether they take discretionary accruals into account while making investment decision or just calculate total accruals and adjust them in their decisions. Finally, it helps in portraying some recommendations for decision-making bodies to safeguard the interests of their organizations investors. Investors have been found so knowledgeable to see through these accruals that which part of reported earnings has been earned actually and which component is outcome of management earning tactics. Previous studies on firm value and discretionary accruals shed light on the issue that firm specific characteristics may also have significant effect and sometimes may wrongly attribute firm value with discretionary accruals. Therefore, bank specific economic characteristics have also been used along with accruals to estimate their effect on value.

Review of Literature
Two schools of thoughts exist in literature on the topic of earning management using artificial smoothing techniques (abnormal accruals). The first school of thought is of the view that earning management techniques reduce the variability of earnings and, therefore, shareholders benefit because the reduced uncertainty and improved predictability of future earnings help in enhancing price-earning multiples. Beidleman [1973] and Lipe [1990] are main advocates of this view; however, they claim that abnormal accruals over time tend to reverse and are readily detected by investors.

447

Zaheer Abbas and M. Faisal Rizwan

The second and more holistic view is that manipulating income in contravention of sound accounting practices adversely affects the shareholder value and in contrary to an underlying concept of representational faithfulness of earnings. Schipper and Vincent [2003] were strong advocates of this view. Closely associated with earning management is the concept of earning quality, which according to Schipper and Vincent, is the extent to which reported earnings faithfully correspond to change in net economic assets other than transactions with owners. Their concept is a departure from the earnings quality constructs based on time series properties of earnings (i.e. persistence, predictability, and variability of earnings). However, both schools of thought believe that managers use discretionary accruals to convey their private information to investors. Dechow et al [1995] evaluated the relative performance of competing models for measuring discretionary accrual and showed that all the models appear well specified when applied to a random sample of firm year and concluded that Jones Model [1991], modified to detect revenue based earning management, provides the most powerful test of earning management and generates fewest type II errors.1 Guay, Kothari, and Watts [1996] pointed out that comparison of discretionary accrual models in Dechow, Sloan, and Sweeney [1995] is based on weak assumptions e.g. evaluation of discretionary accruals models using stock returns. Schipper [1989] provides a conceptual framework for analyzing earnings management from an informational perspective. Deangelo [1988] refers to earnings management in buyout cases. Teoh, Welch and Wong [1998a, 1998b] find that firms manage earnings prior to seasoned equity offers and IPOs. Burgstahler and Eames [1998] conclude that firms manage earnings to meet financial analysts forecasts. Dye [1988] and Verrechia [1986] propose analytic models of earnings management. Stolowy and Breton [2000] suggest three broad objectives for earnings management: minimization of political costs; minimization of the cost of capital and maximization of managers wealth. Soon Suk Yoon [2005] in A Comparison of Earnings Management between KSE (Korea Stock Exchange) Firms and KOSDAQ showed that KOSDOQ firms tend to more actively manipulate earnings to avoid losses than to KSE (Korea Stock Exchange) firms. Soon Suk Yoon, Gary Miller and Pornsit Jiraporn in Earning Management Vehicles for Korean Firms [2006] showed that there are clear discrepancies in earning management vehicles firms use when firms manage earnings in different directions. Ben Hsien Bao et al [2004] argue that lower variability of earnings does not guarantee income smoothers higher firm values. Instead, smoothers earning should be more value relevant if they are of high quality i.e. earning quality should be considered simultaneously. Sample firms in their study are divided into four groups: quality earning smoothers, quality earning non smoothers, non earning quality smoothers and non quality earning non smoothers. I extend these studies on banking industry and suggest that banks face negative response in market when earning is managed through abnormal accruals. In summary, one school of thought has view that earning management techniques reduce earning variability and therefore shareholders benefit from it and should respond positively. The second view is somewhat holistic view that manipulating earning is in contravention of sound accounting practices and adversely affects shareholder value. According to this view, shareholders should respond negatively. What view or what kind of shareholders response to earning management in banking sector of Pakistan holds true has not yet been addressed empirically till now. Whether models available in literature are successful or not in decomposing the total accruals into discretionary and non-discretionary accruals is still an open question. Different researchers have shown relationship between firm value and accrual management in different countries but no such research has been conducted in Pakistan. Earning management has become buzzword among academic circles but whether it has positive, negative or no effect on firm value is open question, which has been addressed in detail in this study.
1

Accepting null hypothesis when it should be rejected

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan

448

Methodology
This chapter presents a complete account of all matters related to the conduct of the study. Measuring Total Accruals (1) Where: TAt is total accruals in year t N.It is Net Income in year t CFOt is cash flows from operating activities in year t

Estimation of Discretionary Accruals


The models to measure discretionary accruals range from the simple, in which total accruals have been used as a measure of discretionary accruals to relatively sophisticated regressions, which decompose total accruals into discretionary and non discretionary accruals. According to cross sectional Modified Jones Model

(2) Where NDAt is non-discretionary accruals in year t scaled by lagged total assets REV is revenues in year t less revenue in year t-1 PPEt is gross property plant and equipment at the end of year t; At-1 is total assets at the end of year t-1; and 1,2,3 are firm specific parameters Estimates of the firm specific characteristics 1,2,3 are obtained by using the following model in the estimation period: (3) Where: TAt is total accruals in year t scaled by lagged total assets is the residual, which represents the firm specific discretionary portion of total accruals while other variables are same as in equation (II).

Capturing firm Value


In line with prior research of Billet et al. [1995], Nohel and Tarhan [1998] and many others, I capture firm value by Tobins Q. Tobin's q, is the ratio of the market value of a firm's assets to the replacement cost of the firm's assets [Tobin 1969]. If a firm is worth more than its value based on what it would cost to rebuild it, then excess profits are being earned. These profits are above and beyond the level that is necessary to keep the firm in the industry. The advantage of using Tobin's q is that the difficult problem of estimating rate of return is avoided. On the other hand, for q to be meaningful, one needs accurate measures of both the market value and replacement cost of a firm's assets.It is usually possible to get an accurate estimate for the market value of a firm's assets by summing the values of the securities that a firm has issued, such as stocks and bonds. It is much more difficult to obtain an estimate of the replacement costs of its assets, unless markets for used equipment exist. Moreover, expenditures on advertising and research and development create intangible assets that may be hard to value. Typically, researchers who construct

449

Zaheer Abbas and M. Faisal Rizwan

Tobin's q ignore the replacement costs of these intangible assets in their calculations. For that reason, q typically exceeds 1. Tobins Q= ((M.V of Equity + B.V of assets)-B.V of Equity)/B.V of Assets (4)

Estimation through Ordinary Least Square Method


This session of thesis discusses the regression of Tobins Q on discretionary accruals, total accruals and non-discretionary accruals. Comparison of all the three regressions has been presented in Table 1. Our null hypothesis is H0: Firm Value decreases if company uses discretionary accruals to hide their earnings in good times and realize it in bad times H1: Firm Value has no relationship with discretionary accruals Discretionary accruals, total accruals and non discretionary accruals have been run as explanatory variables along with other Economic factors in equation V, VI and VII respectively (5) (6) (7) Where TQt is Tobins q measured as the ratio of market value of assets to book value of assets in year t DAt are discretionary accruals in year t NDAt are non-discretionary accruals in year t TAt is total accruals in year t PROFt is profitability in year t, measured as return on assets FCt is financial constraints measured by dummy variable Dividend. If bank pays dividend, it gains value of 1 otherwise 0. Sizet is size of bank measured as natural log of total assets and t is error term.

Data Issues and Sources


To apply Cross Sectional Modified Jones Model [1995] (Dechow, Sloan, and Sweeney), data for all 21 banks has been obtained over period of 2 years ranging i.e. 2003 & 2004. For 2002, only total assets of banks were calculated, which were used for scaling purpose. Different Data has been collected from different sources. However, major chunk was obtained from annual reports of the banks. Further, in summarized form, accounting data of banks has also been obtained from analysis reports presented on website of Karachi Stock Exchange. In chapter form data regarding cash flow positions of banks has been obtained from website of State Bank of Pakistan as it is not available in analysis reports presented on KSE (Karachi Stock Exchange) website. Our data is Pooled data, which has been arranged in stacked form, where all of the data for a variable are grouped together, but separated from data for other variables. Here, the data for different cross-sections are stacked on top of one another, with each column representing a variable. So our data is stacked by cross sectional units i.e. banks. Time effect on intercept and slope coefficients were tested using time dummy variable, which assumed the value of one for year 2002 and 0 for year 2003 but coefficient of this dummy variable was found insignificant showing clearly that intercept and slope do not change over time thus supporting the sufficiency of OLS method.

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan

450

Results
As we are concerned with the magnitude of accruals not with direction therefore, absolute values of total accruals, discretionary accruals and non-discretionary accruals were used as explanatory variables. To avoid the problem of multicollinearity among discretionary, non-discretionary and total accruals, separate regressions were run for each kind in V, VI and VII respectively. Regression 1 TQt = 1 + 2 ( DAt ) + 3 ( PROFt ) + 4 ( FC t ) + 5 ( Sizet ) + t Regression 2 TQt = 1 + 2 (TAt ) + 3 ( PROFt ) + 4 ( FC t ) + 5 ( Sizet ) + t Regression 3 TQt = 1 + 2 ( NDAt ) + 3 ( PROFt ) + 4 ( FC t ) + 5 ( Sizet ) + t
Table 1:
DA TA NDA FC Size PROF Constant
Note: P-Values in parenthesis

Regression comparison
Regression 1 -0.032474 (0.01929) Regression 2 -0.008578 (0.6460) -0.016654 (0.2775) 0.002312 (0.7198) 4.9322 (0.0000) 1.01632 (0.0000) -0.014920 (0.3379) 0.002613 (0.6938) 4.8887 (0.0000) 1.02264 (0.0000) 0.021096 (0.6585) -0.013275 (0.3973) 0.001718 (0.7968) 4.8708 (0.0000) 0.0000 (0.0000) Regression 3

Where DA is discretionary component of total accruals TA is total accruals measured as the difference between net income and operating cash flow NDA is non-discretionary component of accruals scaled by one year and measured as sum of change in revenue minus change in receivable plus operating fixed asset in year t FC is measurement of financial constraints. It represents dummy variable dividend Size has been measured as natural log of total assets; a bank has in its balance sheet at the end of year t Prof is profitability, which has been measured as return on assets in year t Using Tobins Q as proxy for firm value in this analysis, negative relationship has been found between firm value and discretionary accruals. Probability for coefficient of discretionary accruals is sufficiently low (0.01929) and thus indicating that discretionary accruals significantly affect firm value and market responds negatively to earnings shown by earning tactics. However, relationship of firm value with total accruals and non-discretionary accruals is insignificant as probability value has been found very high (0.6460) and (0.6585). These results support the findings of Eli Bartov and Ferdinand A. Gul [2000] where in they found negative relation between discretionary accruals and firm value. In the third regression, where in firm value has been shown as dependent variable on non discretionary component; again insignificant relationship of firm value with non discretionary accruals has been found which strengthens our earlier finding that market has knowledge regarding earning quality of firm as it responds negatively and significantly towards discretionary accruals but insignificantly towards non discretionary component

451

Zaheer Abbas and M. Faisal Rizwan

Parameters statistics of equation V in which effect of discretionary accruals has been captured are summarized in Table 2.
Table 2: Statistics Summary of Regression 1
Values 1.01632 -0.032474 4.9322 -0.016654 0.002312 0.95 0.94 0.041526 2.013650

Parameters

1 2 3 4 5

R Square Adjusted R Square Standard Error Durbin Watson Stat

Regression equation: TQt = 1 + 2 ( DAt ) + 3 ( PROFt ) + 4 ( FC t ) + 5 ( Sizet ) + t As Table 2 reports that the values of both R Square and adjusted R Square strongly suggest that model serves its purpose in determining the effect of accruals on firm value. The Durbin-Watson statistic measures the serial correlation in the residuals. As a rule of thumb, if the DW is less than 2, there is evidence of positive serial correlation. The DW statistic in our output is 2.013650, indicating the chance of serial correlation in the residuals is very low.

Results of Restrictions on coefficient Value


For this purpose Wald Coefficient Test has been used, which determines the chances the occurrence of specific value for coefficient It has been applied on coefficient value of discretionary accruals and chances of its occurrences have been determined. While applying Wald Test on coefficient of discretionary accruals, null hypothesis is stated as coefficient value for discretionary accrual is equal to -0.03247. The output of Wald Coefficient test is shown in table 3
Table 3
Null Hypothesis: F-statistic Chi-square Wald Test: C(1)=-0.03247 (Discretionary accruals) 3.14E-08 Probability 3.14E-08 Probability

0.999859 0.999859

High probability values strongly suggest that the null hypothesis that coefficient value of discretionary accruals is -0.03247 should be strongly accepted. One of the basic assumptions of OLS method is that regressors are not mutually correlated. If more than one of them is correlated with other, multicollinearity is said to exist. Logic behind assumption of no multicollinearity is simple that if two or more independent variables are linearly dependent on each other, one of them should be included instead of both. A suggested rule of thumb is that if the pair wise correlation between two regressors is very high, in excess of 0.8, multicollinearity may pose serious problem. Correlation Matrix has been shown in Table 4

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan
Table 4
DA DA Size Prof DIV 1 -0.127 0.2154 -0.088 Correlation Matrix Size -0.127 1 -0.249 0.4108 Prof 0.2154 -0.249 1 0.2047 DIV -0.088 0.4108 0.2047 1

452

In correlation matrix shown in table 4, no two variables have correlation above 0.8. Thus there are very few chances that regressors will contaminate results due to multicollinearity. Another important assumption of OLS method is homoscedasticity, which means that residuals have same variance. Heteroscedasticity may also arise as a result of the presence of outliers. An outlying observation or outlier is an observation that is much different in relation to observation in the sample, which justifies the scaling of discretionary accruals and Rev-Rec and PPE by lagged total. Just like multicollinearity, there is no hard and fast rule for detecting heteroscedasticity, only a few rules of thumb

White's Heteroscedasticity Test


This is a test for heteroscedasticity in the residuals from a least squares regression [White, 1980]. Whites test is a test of the null hypothesis of no heteroscedasticity against heteroscedasticity of some unknown general form. Output of White Heteroscedasticity test has been shown in table 5
Table 5
F-statistic Obs*R-squared White Heteroscedasticity Test: 1.404245 Probability 9.419363 Probability 0.235695 0.223933

TQt = 1 + 2 ( DAt ) + 3 ( PROFt ) + 4 ( FC t ) + 5 ( Sizet ) + t The Obs*R-squared statistic is Whites test statistic. The probability value of Obs *R Squared is sufficiently high thus we cant reject the null hypothesis of no heteroscedasticity. Thus the absence of multicollinearity and heteroscedasticity support the use of OLS method in our estimation.

Conclusions
In this thesis, commercial banks listed at Karachi Stock Exchange have been investigated using their panel data from 2003 to 2004. The role of discretionary accruals in locating a firm at a specific value has been empirically analyzed. First of all total accruals have been calculated using cash flows statement based approach. Then Discretionary accruals have been estimated using cross section modified Jones model [1995]. Similar to Billet et al [1995], Nohel and Tarhan [1998] and many others, Tobins Q has been used as proxy for firm value. To capture firm specific characteristics, size, profitability and financial constraints of firms have been used along with discretionary accruals in this thesis as explanatory variables. Our findings suggest that if management wants to increase firm value, it must focus on earnings quality instead of earning tactics such as discretionary accruals as empirical results have found that market responds negatively towards these discretionary components thus resulting in decrease in value. Our findings prove that conclusion of Yan Zhang, Pingshun Huang, Donald R. Deis and Jacquelyn Sue Moffit holds true in Pakistan and management has to face problem of share prices if it deliberately misguides the market participants as they can see through the company and respond accordingly. Thus higher officials of the firms should focus on real story of earnings otherwise; firm has to face value

453

Zaheer Abbas and M. Faisal Rizwan

problem in the market. The second finding, which is significant in all the three regression models, is relationship of firm value with profitability, which is very strong positive relationship. In firm specific economic characteristics used in this model, however, only profitability has significant impact on value. Market value has strong relationship with profitability of company and is significant in all the three regressions. So high profits serve in locating a firm at higher value but decomposition of profit into discretionary and non-discretionary components determine the exact position for a firm. The major contribution of this study is the extent of awareness issue, the extent to which market participants go into company accounting figures, available publicly. Empirical results of this study revealed that earning management is being practiced in banking industry as mean difference between two groups of banks with above mean discretionary accruals and below mean discretionary accruals has been found significant. This paper may suffer from some limitations. This may be early to say that firm value has insignificant relationship with total accruals. This finding may need further analysis. Another limitation is that the discretionary accruals approach is subject to misspecification problems. However, this misspecification is not unique to our study as most of the earning management studies use variety of discretionary accruals approaches. Secondly, this study assumes that there is strong relationship between earning figures made publicly available and market reaction. This study also provides framework for further research in the area of earning management. For example, there is still open question that which of the discretionary accruals model measures the discretionary accruals more accurately. Secondly, relationship of earning management with audit qualification is also the area where an independent research may be undertaken.

Abnormal Accruals and Firm Value, Panel Data Analysis of Banking Industry of Pakistan

454

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] Lambert R.A [1984] Income Smoothing As Rational Behavior The Accounting Review 59, 604-618. Billet , M. Flannery, J. Garfinkel [1995] The Effect of Lender Identity on Borrowing Firms Equity Returns Journal of Finance 50, 669-718. Nohel, T. and V Tahran [1998] Share Repurchases and Firm Performance: New Evidence on Agency Cost Of Free Cash Flow Journal of Financial Economics 49, 187-222. Dechow, P.M., R.G Sloan, and A.P Sweeney [1995] Causes and Consequences Of Earning Management: An Analysis Of Firms Subject Enforcement Actions By The SEC Contemporary Accounting Research 13, 1-36. Beidleman, C.R [1973] Income Smoothing, The Role of Management The Accounting Review 48, 653-667 Lipe, R [1990] The Relation between Stock Returns and Accounting Earnings Given Alternative Information The Accounting Review 65,49-71 Schipper, K and L Vincent [2003] Earning Quality Accounting Horizons 17, 97-110 Dechow, P.M., R.G. Sloan, A.P. Sweeney [1995] Detecting Earning Management The Accounting Review 70, 115-139 Guay W. and S.P. Kothari [2003] How Much Do Firms Hedge With Derivatives? Journal of Financial Economics 70, 423-461 Menon, K and D. Williams [2004] Former Audit Partners and Abnormal Accruals The Accounting Review 79, 1095-1118 Xie, B., W.N Davidson, P.J DaDalt [2003] Earning Management and Corporate Governance: The Role of the Board and the Audit Committee Journal of Corporate Finance 9, 295-316 Klein, A. [2002] Audit Committee, Board of Director Characteristics, And Earning Management Journal of Accounting and Economics 33, 375-400 Levitt, A [1998] The Numbers Game Remarks Delivered At the NYU Center for Law And Business, New York Pincus, M.A and S.R. Thiagarajan [2000] Risk Measurement and Hedging: With and Without Derivatives Financial Management 29, 5-30 Pincus, M. and S. Rajgopal [2002] The Interaction of Accrual Management and Hedging: Evidence from Oil and Gas Firms The Accounting Review 71, 27-160 Smith, C. and R Stulz [1985] The Determinants of Hedging Policies Journal Of Financial And Quantitative Analysis 20, 391-405 Visvanathan, G. [1998] Who Uses Interest Rate Swaps? A Cross-Sectional Analysis Journal of Accounting, Auditing And Finance 13, 173-200 Yan Zhang, Pinghsun Huang, Donald R Deis, Jacquelyn Sue Moffitt, [2005] Discretionary Accruals, Hedging and Firm Value Financial Management Association, Annual Meeting Louisiana State University

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.455-476 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Static Analysis of Slip Damping with Clamped Laminated Beams


O.Damisa Faculty of Engineering, University of Lagos Department of Mechanical Engineering V.O.S. Olunloyo Faculty of Engineering, University of Lagos Department of Systems Engineering E-mail: vosolunloyo@hotmail.com C.A. Osheku Faculty of Engineering, University of Lagos Department of Mechanical Engineering A.A. Oyediran AYT Research Corp., McLean VA. 22102, USA Abstract The use of static analysis of clamped laminated beams as a model for investigating noise dissipation in turbine blades for supersonic and advanced subsonic engines is herein revisited as a model for examining vibration energy dissipation by interfacial friction in jointed structures. This derives from the earlier work of Pian and Hallowell, Goodman and Klumpp, and more recently Damisa. Based on heuristic arguments and experimental results, Goodman and Klumpp postulated a parabolic profile for the energy dissipated by clamped laminated cantilever beams under constant pressure. By invoking operational methods and carrying out a formal analysis, this paper derives explicit expressions for the relative slip and energy dissipation, which are shown to be in close agreement with the experimental results reported by Goodman and Klumpp. Furthermore, the analysis presented here provides for variation of clamping pressure along the beam and our results are applicable over a wider range of interface pressure than what was studied by earlier workers. In particular, it is shown that linear variation in interfacial pressure distribution can act to modulate the energy dissipated. Keywords: laminated beams; slip; energy dissipation; optimal pressure for energy dissipation

1.0. Introduction
In the early 90s, two types of engines were proposed for aircraft power plants viz: supersonic and advanced subsonic engines with advanced technology. In both cases, the exit temperature requirement is in the neighbourhood of 2000C. At such a high temperature, most turbine blades will melt. Consequently, research effort was focused on the use of ceramic materials such as hybrid of ceramic

Static Analysis of Slip Damping With Clamped Laminated Beams

456

with titanium and tantalite for the turbine blades to withstand the required exit temperature. To be sure, service failures of gas turbine blades can, in many cases, be attributed to the high stresses accompanying resonant vibration. These stresses are known to reduce the fatigue life of the blades in gas turbines. However, against the background of economic considerations, attention was also given to composite beams with the upper laminate made of ceramics that can withstand the exit temperature whilst the weaker lower laminate can be of relatively cheaper material. Such innovative proposals sparked off investigations into the vibration characteristics and distribution of in-plane and out-plane shear stresses through the turbine blades. In an earlier effort, a theory of structural damping in a built-up beam had been described by Pian and Hallowell [1] who considered a beam fabricated in two parts and connected together by riveted cover plates. However, their analysis is limited by idealized properties attributed to the riveted connections and the cover plates. Goodman and Klumpp [2] on the other hand examined the energy dissipation by slip at the interface of a laminated beam. Theirs is regarded as a first-order theory predicated on static conditions as the time-variation or frequency-dependence was not considered. Recently, Damisa [3] has derived expressions for in-plane and out-plane stresses through a laminated beam. His formulation was based on the constitutive equations of elasticity and the dynamic equations of motion, taken along with the usual beam reductions. By doing this, he was able to arrive at expressions for the magnitudes of the slip and energy dissipation for general support conditions. This was specifically applied to the cantilever beam, and he thereafter compared his results to Goodman and Klumpps. His formulated governing equation of motion may be regarded as the non-homogeneous Euler-Bernoullis classical beam theory equation with a space derivative of the clamping pressure term. Since his emphasis was only on the analysis of the slip that goes on at the interface, for which the classical theory gives enough information, he reverted to the theory for uniform pressure reported in [2] thus solving Euler-Bernoullis classical beam theory equation for the analysis of the stress distribution, slip damping and energy dissipation at the interface of the Laminated Beam. The report of Goodman and Klumpp [2] also revealed that at favourable conditions the energy dissipated in slip damping may be quite large compared with that dissipated in internal friction and the total strain energy stored in the blade. In particular, the measurements of slip damping reported by earlier investigators were shown to be in close agreement with their results. However, the effects of large amplitude on the relative slip and energy dissipation at the interface of the layered beam was missing in their report. The study of large amplitude vibration of simply supported beams can however be traced to the work of Krieger [4], wherein the governing partial differential equations were reduced to ordinary differential equations and a solution was obtained in terms of elliptic functions using a one-term approximation. Similarly, Burgreen gave the solution for the large amplitude vibration problems of hinged beams based on the classical continuum approach [5]. Theirs were completely dynamical problems which preceded Goodman and Klumpps work, where the analysis was limited to static conditions both in theory and experiment. Damisas analysis can however be regarded as a dynamic rather than static problem and in all no consideration was given to the effects composite materials will impose on the mechanism of slip damping and energy dissipation at the interface. Composite materials are generally regarded as good candidates to replace and complement conventional materials like steel, aluminium or concrete, in view of their relative light weight, corrosion resistance, electromagnetic transparency, etc. However, earlier studies demonstrate that connections performance is one of the more important limiting factors for the application of composites. In fact bolted connections are preferred because they facilitate construction, and can be easily disassembled. As pointed out by Barbero, they also facilitate rehabilitation of damaged structures and are generally considered safer than bonded connection for long life cycles [6]. Experimental evidence as reported by Barbero and noted by Doyle [7] however suggests the possibility that friction plays an important role in the bolted and the bonded-bolted connections strength. In particular, the amount of applied torque is known to influence the ultimate strength of both bolted and bonded-bolted connections, which may indicate that slip occurs at different levels of load.

457

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

The nature of the interface pressure profile across the beam layer is a separate and important issue that has received some attention over the years. Indeed there are several ways of simulating such interfacial pressure including mechanisms such as bonded (welded) connections or the use of bolted connections and even bonded-bolted connections placed at appropriate locations along the laminate interface. Notwithstanding the importance of this issue all previous analytical work have assumed a uniform pressure profile across the interface. However several workers, especially in the machine tool literature have tried to measure the actual nature of the pressure that is generated by investigating various forms of securing the layers. In fact following the work of Goodman and Klumpp, early workers, such as Masuko et al [8], Nishiwaki et al [9,10], Motosh[11] had assumed uniform or constant intensity of pressure distribution at the interface of such layered structures without considering the effects of surface irregularities and asperities. However , subsequent studies from several authors such as Ferlund [12], Greenwood [13], Lardner [14], Bradly [15], Sidorov [16], El-Zabry [17], Kaboyashi and Matsubayashi [18,19], Tsai et al [20], Shin et al [21] and Song et al [22] reported on the actual experimental values obtained for the pattern and intensity of pressure distribution at the interface of bolted joints and presented the corresponding damping characteristics for plates and shells. In particular, Gould and Mikic[23] ; Ziada and Abd [24] reported that the pressure distribution at the interface of a bolted joint is parabolic and has an influence zone in the form of a circle that is 3.5 times the diameter of the connecting bolt. It was also reported as being independent of the tightening load applied. On the analytical side, Hanson and Spice [25] have investigated the structural damping in laminated beams due to interfacial slip. In their formulation, a model in which slip can occur along the interface was analysed for a two-layered plate under the assumption that an adhesive layer of negligible thickness and mass bonds the two adjoining surfaces such that a restoring force that is proportional to the amount of slip is created by the adhesive medium. More recently Nanda and Behera [26] developed a theoretical expression for the pressure distribution at the interface of a bolted joint by curve fitting the earlier data reported by Ziada and Abd [24]. By postulating a priori an even function, they obtained an eighth order polynomial in terms of the normalised radial distance from the centre of the bolt such that the function assumes its maximum value at the centre of the bolt and decreases monotonically radially. Their result also indicated that apart from the first two terms the coefficients obtained for the polynomial were relatively insignificant. This suggests that for a cantilever beam where interface pressure is effected through use of bolted joints, a linear profile radiating from the clamped end can be reasonably postulated as a first order approximation for the pressure distribution across the interface. To put the previous work in their proper context, this paper deals with static slip and the energy dissipated at the interface of laminated beams. In particular, we investigate the contribution of each half of the composite laminates and the variability or otherwise of the optimum clamping pressure under static conditions. We have for the purpose, derived a time-independent governing ordinary differential equation for the laminated beams to advance our analysis.

2.0. Stress Analysis


Fig. 1.1a shows the geometry and coordinate axes for a layered beam of width b, whilst h is the uniform depth of each layer. The layers are held together by a pressure distribution p(x) applied to both the upper and lower surfaces as illustrated below:

Static Analysis of Slip Damping With Clamped Laminated Beams


Figure 1.1a: Laminated Beam of Similar Material, Coordinate Axes and Geometry

458

p(x) h h (2) (1) h h x

b (2) (1)

2h

p(x) z Consider an element through the layered beam as shown in Fig. 1.1b and the stresses through the element. Fig. 1.1c on the other hand shows all the stresses associated with microscopic slip at the interface for each laminate.
Figure 1.1b: Pre-slip free body diagram

459

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran


Figure 1.1c: Post-Slip free body diagrams

Upper Laminate

Lower Laminate

For each of the laminate elements at equilibrium, dS =0 dx (1) Taking moments about the centre of each of the laminates at equilibrium gives for the upper and lower laminate respectively, dM 1 S = bhp dx 2 (2) so that substitution into eqn (1) gives d 2M 1 dp bh =0 2 2 dx dx (3) 2.1. Governing Equations for the Static Problems If we invoke the relation between bending moment and curvature and the approximate curvaturedisplacement relation, used in determining static deflections of beams as derived by Warburton in [27] viz: d 2W M = EI dx 2 (4) We may then rewrite equation (3) as d 4W 1 dp EI bh =0 4 2 dx dx (5) Hence, for similar materials, under static conditions, the governing differential equation is the same for each laminate.

Static Analysis of Slip Damping With Clamped Laminated Beams

460

3.0. Analysis of Static Response


As evident from equations (4) and (5), the governing equation for the static case admits the form d 4W dp = 4 dx dx where 6 = Eh 2 We now introduce the Fourier Finite Sine Transform namely: L 2 nx nx s {[]} = [] = []sin dx ; [] = []sin L L L n =1 0

(6)

(7)

On invoking Fourier Finite Sine Transform equation (6) admits the form n 3 3 n 4 4 n +1 s { ( x )} 3 W (0 ) + ( 1) W (L ) W 4 L L n dp n +1 + W xx (0 ) + ( 1) W xx (L ) = s L dx (8) where the subscript x denotes derivative with respect to x. On applying the usual cantilever boundary conditions equation (8) can be further expressed as: n 4 4 F n 3 3 n dp n +1 W ( n ) 3 ( 1) W (L ) + Wxx (0) = s 4 L L L dx (9) where n n = L We shall limit our analysis for now to the case of linear pressure variation namely: P( x ) = P0 1 + x L (10) where is the constant slope of the pressure variation. Consequently, we rewrite equation (9) in the form 6P0 F n 4 4 F n 3 3 n n +1 Wxx (0) = W ( n ) 3 ( 1) W (L ) + 1 4 L L L Eh 2 (11) where L nx L n +1 1F = sin 1 + ( 1) dx = L n 0 (12) The analysis of equation (11) can be further simplified by invoking the Goodman and Klumpp end condition namely: h F xz dz = 2b at x = L 0 (13) which merely reflects the application of a transverse force F at the free end of the beam. Furthermore, we use the shear stress relation namely: P 1 ( xz )1 = E (z 2 hz )W xxx (z h) h 2 (14) Hence, in view of equations (10) and (14), equation (13) now admits the form h h h P0 P0 1 F 2 2 E z hz W xxx dz h (z h)dz h (z h )dz = 2b 0 0 0 (15)

461

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

Equation (15) can be further integrated to give: L h L h L h P0 P0 1 2 2 E z hz Wxxx dzdx h (z h )dzdx h (z h)dzdx 0 0 0 0 0 0

2b dx
0

(16)

(17) Substitution of equation (17) into equation (11), leads to the Fourier inversion ( 1)n+1 sin nx 2W (L ) n L n =1 6P0 6P0 1 nx 6F 2 L3 W (x ) = 3 3 sin 3 2 2 L Eh Eh n =1 n Ebh 3 P0 3 1 2nx + L 5 5 sin 2 L 8 Eh n =1 n (18) The evaluation of the series in equation (18) is further simplified if we invoke the well known closed form Fourier series representations namely: sin nx 2 x x 2 x 3 = + , (0 x 2 ) n3 6 4 12 n =1 (19) and 4 x 2 x 3 x 4 x 5 sin nx = + , (0 x 2 ) n5 90 36 48 240 n =1 (20)

Here P0 is the uniform clamping pressure introduced via equation (10). It can be immediately deduced from equation (16) that 6 P0 6 P0 6F W xx (0) = L 3 Eh 2 Eh 2 Ebh

Similarly, we can also invoke the Fourier Series representation of the function f(x) = x in the interval L < x < L namely, n +1 2 L ( 1) x = sin nx (21) n =1 n So that the Fourier series representing the function f ( x ) = x in the interval 0 < x < 1 admits the form n +1 1 ( 1) x = sin nx (22) n =1 n which now allows us to write result (18) in the form 6P0 6P0 x x 2 x 3 3 6F + 2W (L )x 2 L 3 Eh 2 Eh 2 6 4 12 Ebh W (x ) = (23) 3 4 5 x x x 3 P0 3 x L 2 + + 2 45 8 Eh 2 9 3 15 dW (0 ) If we also impose the boundary condition = 0 on equation (23), we then obtain dx P P 3 P0 3 F W (L ) = L3 0 02 L (24) 3 2 Eh Eh (8)(90) Eh 2 Ebh

Static Analysis of Slip Damping With Clamped Laminated Beams as the static deflection at the free end of the clamped laminated cantilever beams. Thus, we can now express equation (23) in the form 3 F P P 0 0 2 x 2 3L 3 Eh 2 Eh Ebh P P F L3 0 0 2 x 3 W (x ) = 3 2 Eh Eh Ebh 3 4 5 3 P0 3 x x x + + 2 L 2 2 8 Eh 9 3 15 which can be further simplified as 11x 3 x 4 x 5 + W = (1 P0 ) 3x 2 x 3 + P0 3x 2 + 12 8 20 where P0 W ( x )Ebh 3 W = , dimensionless response ; and P0 = , dimensionless pressure 3 F LF bh Thus, in the limit as 0 , the static response for constant pressure assumes the form: W = (1 P0 )(3 x 2 x 3 ) This result indicates that there is a critical pressure for which there is no deflection 1 P0 = .

462

(25)

(26)

( )

(27) viz:

4.0. Evaluation of Slip at Static Conditions


The relative slip at the interface of the laminated beams is given by u ( x ) = u 2 ( x,0 ) u1 (x,0 + ) Following Goodman and Klumpp, this can also be written as

(28) (29)

u ( x ) = E 1 {( x )1 ( ,0 ) ( x )2 ( ,0 + )}d
0

By invoking Damisas bending stress relations, we now express equation (29) in the form x d 2W 2P0 ( L ) + 2P0 ( 2 L )d u ( x ) = E 1 h + (30) 2 Eh LEh d 0 which can be non-dimensionalised as x x x u Ebh 2 d 2W = d + 2P0 ( 1)d + 2P0 ( 2 )d (31) L2 F d 2 0 0 0 and integrated to give dW 2 (32) u = + P0 (x 2 2 x ) + P0 x 3 x 2 dx 3 where u Ebh 2 u = L2 F so that substitution for W from equation (27) gives the result 7 7 1 (33) u = (4 P0 3)(x 2 2 x ) + P0 6 x + x 2 + x 3 x 4 4 6 4 Hence, in the limit as 0 , the slip at constant normalized pressure takes the quadratic form:

463
u =

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

(4P

3) x 2 2 x

(34)

which suggests that even in the presence of coulomb friction and interface pressure P0 , there are critical values of pressure (i.e. P0 = 0.75 1 ) for which no slip occurs.

5.0. Energy Dissipation


The energy dissipated per cycle, following Goodman and Klumpp is given by the relation

D = 4b P( x )u ( x )dx
0

(35)

which can also be expressed as D where = P0 1 + 2 0 so that equation (36) is now written as 1 (37) D = 4P0 1 + u dx 2 0 which on substituting for u from equation (33) gives 4 421 2 261 D = 6 P0 8P0 2 + 3P0 P0 P022 (38) 3 40 80 where DEbh 3 D = , is the dimensionless energy dissipated L3 F 2 Hence, in the limit as 0 , the energy dissipated at constant normalized pressure admits the form: 4 D = P0 (6 8P0 ) (39) 3 which is exactly the same equation that was derived by Goodman and Klumpp.[2] On the other hand, by considering equation (38) as a quadratic for P0 we can find the critical value of the interface pressure for which no energy is dissipated viz: 1 421 261 2 (40) + P0 = 0 . 75 1 1 + 1 + 320 640 2 which for the case of uniform interface pressure ( = 0 ) vanishes for the values = 0, or P0 = 0 ; Pav = P0 (1+ x )dx
1

= 4 Pav u dx
0

(36)

0.75-1 with the last case being consistent with the occurrence of absence of slip predicted from equation (34).

6.0. Analysis of Optimum Clamping Pressure


The optimum clamping pressure can be found from the partial derivative of the energy dissipated if we set 421 2 261 D 4 6 P0 8P0 2 + 3P0 = P0 P022 = 0 (41) 40 80 P0 P0 3

Static Analysis of Slip Damping With Clamped Laminated Beams

464

Thus, we derive the general expression for the optimum clamping pressure as 1 421 261 2 (42) + P0 = 0 . 375 1 1 + 1 + 320 640 2 3 as was earlier so that in the limit as 0 , we recover the optimum clamping pressure P0 = 8 reported by Goodman and Klumpp. Furthermore, at this optimal pressure, the energy dissipation can be computed from equation (38) as 3 421 261 Dmax = H 2+ H 1 + H 2 (43) 2 320 640 where

261 2 421 = 1 + 1 + + 640 2 320 261 +2 = 1 + 2 640


1

320 + 261

(44)

261 = 1 + 320 which indicates that even when linear pressure variation is admitted, the maximum dissipated energy still remains independent of the coefficient of friction as was thought by earlier workers such as Earles and Beards [28] and Damisa [3]. However, relation (44) indicates that slip damping assumes its 320 maximum value in the neighbourhood of = 261 Moreover, relation (42) also indicates that for our results to be meaningful, the interface pressure must be positive definite; thus the requirement > 1.226 must hold. However, whether is positive or negative depends on the nature of the clamping and how the clamping pressure is applied but we shall return to this point in the next section.

7.0. Analysis of Material Damping


The bending stress relation satisfies the form 2 ( x )1 (x, z ) = E (2 z h ) W + Pav (x L ) (45) h 2 x 2 Now at z = h, this gives the extreme fiber stress as Eh 2W Pav (46) ( x )1 (x, z ) = + (x L ) 2 x 2 h The maximum value of equation (46) is located at the built-in-end, i.e. x = 0 If we define m , as maximum extreme fiber stress in the absence of friction, then by substituting appropriately for W(x,z) in equation (46), we find ( x )1 (0, h )bh 2 m = = 3(P0 (1+ ) 1) P0 1 + = 3 (47) FL 2 = 0 =0 Since material strain energy is independent of friction, it then satisfies the relation 2 1 ( x )1 (0, h ) U = E 2

(48)

465 where ( x )12 (0, h ) =


2 m F 2 L2

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

b2h4 or in non-dimensionalised form as 1 2 U = m 2 where Ub 2 h 4 U = 2 2 F L Thus, equation (48) indicates energy dissipated through material damping.

8.0. Comparison of Slip and Material Damping


By invoking the relations for slip and material damping energies, we write the comparison ratio as D slip energy = o = material energy U where 18D0 E D0 F 2 L3 18Eb 2 h 4 D0 2 2 2 = (49) = = 2 2 F L U Ebh m ( x )1 (0, h ) which can be arranged in the form, bh 2 L( x )1 (0, h ) 2 18D0 = D0 (50) 2 18E m as was earlier reported by Goodman and Klumpp. This relationship provides a basis for comparison of the relative efficacy of the two mechanisms for dissipating energy and we shall return to this issue in the subsequent section.

9.0. Discussion of Results


The results of our analysis indicate that there are critical values of the normalized interface pressure P0 at which displacement, slip or dissipated energy disappear even when there is coulomb friction as tabulated below:
Objective Normalised Interface Pressure Uniform Pressure P = P0
P = P0 (1+ x ) Pav = P0 1 + 2

No deflection

(W = 0)
1

No slip (u = 0 )

Max energy Dissipation

(D = D )
max

P0 =

P0 = 0.75
1

P0 = 0.375 1 P0 = 0.375 1H

P0 =

(1+ )

P0 = 0.75 1 H

Here

H = 1 + 261 320 The objective in noise dissipation in either turbine blades or jointed structures in general is to have as much energy dissipation with minimal blade or structural deflection. Earlier workers have also observed that interface pressure is rarely uniform. The strategy here then is to find out what type of

Static Analysis of Slip Damping With Clamped Laminated Beams

466

interface pressure distribution allows maximum energy dissipation on the one hand and how do you simulate such a pressure distribution while at the same time maintaining minimal structural deflection as in the case of aerodynamic structures. For the case of uniform interface pressure, Figure 1 shows that the results from our analysis agree with the experimental results reported in [2] by Goodman and Klumpp and that the level of slip at the interface grows as the corresponding applied interface pressure increases. Figure 2 on the other hand shows that the corresponding interface pressure for maximum energy dissipation is always lower than what is required to guarantee absence of slip at the interface of the beams. This in turn is lower than the level of interface pressure that would guarantee that the entire structure suffers no transverse deflection.
Figure 1: Normalised Slip Profile at Uniform Interface Pressure

467

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

Figure 2: Normalized optimum clamping pressure versus Coulomb friction coefficient for case of UNIFORM interface pressure

The results obtained in this study are also of some practical interest. They emphasize the need to pay attention to the characteristics of the interface pressure distribution. Under assumed constant pressure distribution, earlier researchers, for example Earles and Beards [28] and Damisa [3], have stated that slip damping is independent of the coefficient of friction at the interface, and therefore the materials of the beam can be selected for other reasons e.g. minimising fretting, without fear of changes in the coefficient of friction. But this study, based on variable pressure distribution, has shown that the characteristics of the pressure distribution can also influence or modify the amount of energy dissipated. In actual practice, even if a fairly uniform pressure distribution can be achieved, further clamping of the built-in end of a cantilever that is usually used in experiments will superpose greater clamping pressure at the built-in end than at the free end, and such variation can, as in this case, be modelled as linear variation. When linear variation of pressure is allowed we find that the response of the structure is strongly influenced by the nature of the variation. In fact there are distinct differences in response when the interface pressure is allowed to decrease (i.e. < 0 ) or increases (i.e. > 0 ) linearly from the clamped end of the beam. For example we observe from Figures 3 & 4 respectively that the deflection of the beam and the corresponding slip at the interface are comparatively lower than what obtains for uniform interface pressure when the slope > 0 whereas they lie above the uniform interface pressure curve when < 0 . Of course, in all cases irrespective of the value of , the overall deflection or slip grows continuously from the clamped to the cantilever end of the beam reflecting the bending effect of the force F applied at the free end.

Static Analysis of Slip Damping With Clamped Laminated Beams


Figure 3: Normalized Blade deflection at optimum clamping pressure for energy dissipation

468

Figure 4: Normalized slip at optimum clamping pressure for energy dissipation

469

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran

Since energy dissipation is the major objective of our analysis, we wish to examine how this is affected by the type of pressure variation allowed at the beam interface. In this regard, figure 5 shows that in all cases, the optimal pressure for maximum energy dissipation varies with the value of with the smallest values associated with positive values of and the highest values occurring for negative values of .
Figure 5: Normalized optimum clamping pressure versus Coulomb friction coefficient for case of LINEARLY varying interface pressure

However when < 0 , there are restrictions on its permissible range since the optimal clamping pressure, equation (42), must always be positive. Thus we derive the condition 261 1+ > 0 ; when 320 To illustrate this point, we plot in figure 6 the energy dissipation profile for the case = 0.14 earlier studied by Goodman & Klumpp [2]. In particular, we note here that as decreases, the optimal interface pressure shifts to the right and properly recovers Goodman and Klumpps value of 0.375 for = 0 . The relative ordering of the curves with respect to decreasing values of in this case is characteristic as we shall discover for other values of later.

Static Analysis of Slip Damping With Clamped Laminated Beams


Figure 6: Normalized energy dissipation versus clamping pressure

470

If on the other hand, we compute the maximum dissipated energy that can be generated at the interface and plot it for the various as given by equation (43), we obtain the graph in figure 7 which indicates that although the maximum energy dissipated when you have uniform interface pressure is independent of the friction coefficient as well as of the ratio of the clamping pressure to the applied transverse stress at the cantilever free end, the same cannot be said for the case when the interface pressure is no longer uniform. In fact, for the case of linear variation in pressure distribution simulated 320 by the slope, we find maximum energy dissipation in the neighbourhood = as shown in Figure 261 7. This has practical significance in the design of such structures.

471

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran


Figure 7: Maximum dissipated energy profile as a function of interfacial pressure gradient

The issue here then is, having seen the pattern of maximum energy dissipation, to what extent can you influence the optimal interfacial pressure by an appropriate choice of the pressure gradient within the permissible range? As derived in equation (43), such maximum energy is modulated by the scale function H and this issue is addressed in figure 8 where we have plotted the scaling factor H as a function of the interfacial pressure gradient coefficient . In this respect, we can observe that for < 0 the energy dissipated outstrips those of uniform pressure while the converse is true for cases of > 0 . However, the function H must be positive definite if the laminates are to remain together as one piece.

Static Analysis of Slip Damping With Clamped Laminated Beams


Figure 8: Scale factor for optimal interfacial pressure gradient

472

Finally, to demonstrate that even in cases of static loading, such as this, slip damping can be a more effective mechanism than material damping, we have plotted in figure 9, the relative magnitudes of these mechanisms for dissipating energy for our model problem.

473

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran


Figure 9: Comparison of Slip and Material Damping

Static Analysis of Slip Damping With Clamped Laminated Beams

474

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] Pian, T.H.H., and Hallowell, F.C. (1950): Investigation of Structural Damping in Simple BuiltUp Beams, Technical Report, Contract N5ori-07833 to ONR, Aeroelastic and Structures Laboratory, Massachusetts Institute of Technology, Cambridge, Mass. Goodman, L.E., and Klumpp, J.H. (1956): Analysis of Slip Damping With Reference to Turbine Blade Vibration, J. Appl. Mech., 23, 421. Damisa, O. (2002): Dynamic Analysis of Slip Damping with Reference to Elastic Beams, J. Engrg. Research (JER) 10, nos. 1 & 2; 1 14. Krieger, S.W. (1950): The effect of an axial force on the vibration of hinged bars, J. Appl. Mech., ASME 17, 35-36. Burgreen, D. (1951): Free vibrations of a pin-ended column with constant distance between pin ends, J. Appl. Mech., ASME 18, 135-139. Barbero, E.J. (1991): 3-D finite element modelling of laminated composites by layer-wise constant shear elements, NDE 10, Enhanced Analysis Techniques for Comp. Mater., ASME, 229-237. Doyle, J.R. (1991): Behavior of bolt and adhesive connections in glass fiber reinforced plastic members, M.Sc. Thesis, West Virginia University. Masuko, M., Ito, Y., and Yoshida, K. (1973): Theoretical analysis for a damping ratio of jointed cantibeam. Bulletin of JSME 16, 1421-1432 . Nishiwaki, N., Masuko, M., Ito, Y., and Okumura, I. (1978): A study on damping capacity of a jointed cantilever beam (1st Report; Experimental Results). Bulletin of JSME 21, 524-531. Nishiwaki, N., Masuko, M., Ito, Y., and Okumura, I. (1980): A study on damping capacity of a jointed cantilever beam (2nd Report; comparison between theoretical and experimental results). Bulletin of JSME 23, 469-475 . Motosh M. (1975): Stress distribution in Joints of bolted or riveted connections, Transactions of ASME. Journal of Engineering for Industry. Ferlund, I. (1961): Method to calculate the pressure between bolted or riveted plates, Transactions of Chalmers University of Technology, Gothenberg, Sweden, 245. Greenwood, J.A. (1964): The elastic stress produced in the mid plane at a slab by pressure applied symmetrically at its surface, Proceedings of the Cambridge Philosophical Society, Cambridge, 60, 159-169. Lardner, T.J. (1965): Stress analysis for thermal contact resistance across bolted joints, Transactions of ASME, Journal of Applied Mechanics, 32, 458-459. Bradly, T.L. (1968): Stress analysis for thermal contact resistance across bolted joints, M.S. Thesis, Massachusetts Institute of Technology, Cambridge, MA. Sidorov, O.T. (1983): Change of the damping of vibrations in the course of operation in dependence on the parameters of bolted joints, Strength of Materials, 14, 671-674. El-Zahry, R.M. (1985): Investigation of the vibration behaviour of pre-loaded bolted joints, Dirasat-Eng. Technology, 12, 201-223. Kaboyashi, T. and Masubayashi, T. (1986): Consideration on the improvement of the stiffness of bolted joints in machine tools; Transaction of the Japan Society of Mechanical Engineering 52; 1092-1096. Kaboyashi, T. and Masubayashi, T. (1986): Consideration on the improvement of the stiffness of bolted joints in machine tools; Bulletin of JSME 29; 3934-3937. Tsai, J.S.; Chou, Y.F. (1988): Modelling of dynamic characteristics of two-bolted joints; Journal of Chinese Institute of Engineering 11; 235-245. Shin, Y.S., Iverson, J.C., and Kim, K.S. (1991): Experimental studies on damping characteristics of bolted joints for plates and shells, Trans ASME, J. Pressure Vessel Technology 113, 402-408. Song, S.; Park, C.; Moran, K.P.; and Lee, S. (1992): Contact area of bolted joints interface. Analytical, finite element modelling and experimental study; ASME, EEP, 3; 73-81.

475 [23] [24] [25] [26] [27] [28]

O.Damisa, V.O.S. Olunloyo, C.A. Osheku and A.A. Oyediran Gould, H.H., and Mikic, B.B. (1972): Areas of contact and pressure distribution in bolted joints, Transactions of ASME, Journal of Engineering for Industry 1972, 864-870. Ziada, H.H., Abd, A.K. (1980): Load pressure distribution and contact areas in bolted joints, Inst. of Engineers (India), 61, 93-100. Hansen S.W; Spies R. (1997): Structural damping in laminated beams due to interfacial slip, Journal of Sound and Vibration; 204, 183 - 202. Nanda, B.K., Behera, A.K. (1999): Study on Damping in Layered and Jointed Structures with Uniform Pressure Distribution at the Interfaces, J. Sound and Vibration, 226(4), 607-624. Warburton, G.B. (1976): The Dynamical Behaviour of Structures, 2nd ed., Pergamon Press Ltd., Oxford, England, 204-205, 230-231. Earles S.W.E. and Beards C.F. (1970): Some aspects of frictional damping as applied to vibrating beams, Int. J. Machine Tool Design and Research, 10, 123-131.

Table of Notations
d dx x E I W b h

differential operator space coordinate modulus of rigidity moment of inertia static response width of Laminated beam depth of Laminated beam coefficient of friction at the interface of the laminated beams clamping pressure at the interface of the laminated beams applied end force amplitude length of Laminated beam shear stress at the interface of the laminated beams bending stress at the upper half of the laminates bending stress at the lower half of the laminates displacement of the upper laminate displacement of the lower laminate dummy variable static response in Fourier plane

P F L

(x)2 (x)1 u2 u1 WF

xz

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.476-491 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Ontology-Based Framework for Integration in Ubiquitous Computing Environment


Eduard Babkin LITIS laboratory, INSA de Rouen. Rouen, France State University Higher School of Economics. Nizhniy Novgorod, Russia E-mail: babkin@hse.nnov.ru Habib Abdulrab LITIS laboratory, INSA de Rouen. Rouen, France E-mail: abdulrab@insa-rouen.fr Abstract This article considers integration of sensors networks in Ubiquitous Computing Environment (UCE), and proposes to augment general-purpose middleware solutions for sensors networks by reconfigurable and highly customizable hardware or software components, which are capable of manipulating meta-data in the form of ontologies. Such components, called Ontology Mediators, perform seamless fusion of sensors data and their representation in a form of real-world concepts and relationships for purposes of a certain single user or a small users group. We offer description of a model-driven design methodology and architecture of a development framework. The framework includes a software library called T-Engine with reusable software components for semantic transformations, and extendable hardware components. For practical implementation of the proposed framework it is suggested to explore the graph-oriented ontology transformation approach and the RDF-based representation of ontology in the form of semi-structured data models. Keywords: distributed systems, integration, semantic interoperability, model-driven approach.

1. Introduction
Miniaturization, reduced costs of electronic components and advanced information technologies now open up possibilities to design, develop and deploy thousands of the coin-sized sensors and mechanical devices into multiple locations. This kind of software-hardware systems, pervasively available to the user in every day activities, is named Ubiquitous Computing Environment (UCE) [1, 2], or even Ubiquitous Smart Space [3, 4]. Establishing ad hoc communication via wireless media numerous elements of UCE provide the user with real-time global sensing, context-aware informational retrieval, and enhanced visualization capabilities. In effect, they give extremely new abilities to look at and interact with our habitat. Many researches made a contribution to developing of Sensors and Actuators Networks (SANET), which became a foundation of UCE. There are tiny hardware devices available in practice for building SANET, embedded operating systems, wireless network protocols, and algorithms of effective energy management [7, 8, 9, 10]. Now researchers community demonstrates growing

477

Eduard Babkin and Habib Abdulrab

interest to resolving the next important problem that will be faced by the developers and the users of UCE since a short time. That is the problem of semantic interoperability in the joint context of SANET, existing IT-infrastructure and people society. Resent results [6, 11-15] show applicability of the middleware paradigm for the solution of that problem, and provide for approaches facilitating integration of SANET on the application level of enterprise systems. However in the case of actual wide-area UCE, multiple SANETs spread across administrative borders, enterprises, and even social cultures. Deep involving of tiny computing devices in our everyday activities requires closer coincidence of computer interfaces with peoples way of perception and mental world models. As activities and social experience are different, the mental world models also differ. So, interfacing with the same sensors can be absolutely dissimilar in respect with the style, modality and informational contents. The same raw data collected by sensors can be interpreted differently and can be applied in absolutely divergent contexts. This simple fact breaks a closed world assumption, and requires shedding light of researchers attention on such issues as explicit meta-data representation, and formal modeling of system properties and interfaces to achieve semantic interoperability. In our research we explorer ways to extend known partial middleware solutions in UCE with a consistent model-driven methodology for semi-automated design and development of components called Ontology Mediators. The main purpose of Ontology Mediator is communication with SANETs, data integration and seamless fusion of diverging real-world concepts and relationships in accordance with information needs of certain single user or a small users group. Depending on specific conditions and requirements, implementation of Ontology Mediator varies from specialized middleware components to reconfigurable hardware devices. Tailored for local ad hoc requirements Ontology Mediator provides strong support for the claim [22] that computer products now eventually progress from large, general-purpose, impersonal static forms to portable, personal, flexible, market-targeted forms. Personalization, flexibility, and quick time to market dictates a quickturn design approach. During domain modeling, design, and development of Ontology Mediators different end-user tools, component libraries and algorithms should be used. No doubt, the best results can be achieved when all these elements are combined into the vertical framework, supporting all stages of the methodology. We have designed architecture of such semantic interoperability framework, suitable for loosely coupled distributed systems like SANETs, and developed several software and hardware prototypes to evaluate benefits and afford proof of Ontology Mediator and the design methodology. This framework supports semi-automated design and development of Ontology Mediators, as well as it allows for designing and establishing coherent information flow between different components on the basis of coordinated ontology transformation activities. In the course of the prototype implementation we applied RDF-model transformations in order to provide semantic interoperability and semantic validation, and explored different kinds of software technologies (the JavaSpace, JMS Messaging, CORBA). Altogether, these contributions are used for rapid development of highly customized Ontology Mediators in UCE. In this paper we describe most important characteristics of the proposed framework as follows. Section 2 gives a short description of foundations and relevant topics to our research. Section 3 contains explanation of major steps in our methodology of Ontology Mediators design. The general architecture of the supporting framework is presented in Section 4. Sections 5 and 6 give a detailed view on the most important components of the framework: the Transformation Engine (T-Engine) and the extensible hardware platform. We summarize obtained results and compare them with other known approaches in the conclusion (Section 7). Section 8 presents authors acknowledges and the references are listed in Section 9.

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

478

2. Formal Foundations
From a conceptual point of view, our research is related with knowledge engineering and knowledge integration techniques specialized for pervasive computations [18, 19, 20]. In this area the concept of ontology plays now the leading role for knowledge representation. It is became a good breeding referring to the Grubers pioneer definition of ontology as a specification of a conceptualization [25]. According to [23] ontology serves for strong support in detailed study of all potentially possible entities and their interrelations in some domain of discourse shared by multiple communities; ontology also enables conceptualization and forming categories of the entities committed by those communities. This direct connection of the ontology technique to integration is pointed out by Y. Kalfoglou: An ontology is an explicit representation of a shared understanding of the important concepts in some domain of interest. [24]. In [26] one can see a good collection of more recent cross-references, all of them underline ontology support for achieving interoperability: Ontology can be seen as the study of the organization and the nature of the world independently of the form of our knowledge about it. To build the formal foundations for our methodology of Ontology Mediators design we apply a well-defined and elegant mathematical theory of ontology developed at the University of Karlsruhe [27]. That theory defines a core ontology (the intentional aspect of the domain of discourse) as a mathematical structure S = C, C, R, , R , where C is a set of concept identifiers (concepts for short); R is a set of relations identifiers (relations for short); C is a partial order on C, called concept hierarchy or taxonomy; a function R CC called signature, such that (r)= dom(r), ran(r), where r R, domain dom(r), range ran(r); R is a partial order on R, called relation hierarchy, such that r1 R r2 if and only if dom(r1) C dom(r2) and ran(r1) C ran(r2). Domain-specific dependencies of concepts and relations in S are formulated by a certain logical language (e.g. first-order predicate calculus) which fits a rather generic definition: Let L be a logical language. An L-axiom system for a core ontology is a pair A = <AI, >,where AI is a set of axioms identifiers; is a mapping AI L. The elements of A are called axioms. Extensional definition of the domain of discourse (assertions or facts about instances and relations) is given by description of the knowledge base KB = C, R, I, C, R , where C is a set of concepts; R is a set of relations; I is a set of instance identifiers (instances for short); C is a function C P(I) called concept instantiation; R is a function C P(I2)called relation instantiation; it has such properties: r R, R(r) C (dom(r)) C (ran(r)). The theory provides also names for concepts and relations calling them signs, and defines a lexicon for ontology Lex = GC, GR, GI, RefC, RefR, RefI , where GC is a set of concepts signs; GR is a set of relations signs; GI is a set of instances signs; RefC is a relation RefC GCC called lexical reference for concepts; RefR is a relation RefR GRR called lexical reference for relations; RefI is a relation RefI GII called lexical reference for instances. In summary, a complete ontology O is defined through the structure O = S, A, KB, Lex where S is a core ontology; A is the L-axiom system; KB is a knowledge base; Lex is a lexicon. Such strict mathematical theory of ontology along with other similar approaches formed a solid foundation for machine-readable representation of ontologies in modern information systems and facilitated their practical applications. For example, the research line known as Semantic Web pays great attention to various practical aspects of ontology manipulation for information heterogeneity resolution in the Internet. Examples of XML language application in coordination frameworks [31, 32] illustrate that Semantic Web solutions can be successfully adopted and applied in many different domains. It seems for us that potential usefulness of the Resource Description Framework (RDF) and the RDF Scheme (RDFS) standards [35], proposed initially for semantic enrichment of WEB contents, is much greater. In order to achieve semantic interoperability for UCE we suggest applying RDF standards as the basic technique for the interoperable description of the knowledge base KB. Fusion of the formal theory of ontology together with RDF allows representing the knowledge base in the single frame of semi-structured data models. In [28] semi-structured data are defined as data that is neither raw, nor strictly typed as in conventional database systems. A convenient approach to represent semi-

479

Eduard Babkin and Habib Abdulrab

structured data uses such edge-labeled graph structures which contain both type definition and actual data elements. Describing this case, P. Buneman says about blurring the distinction between schema and instance and proposes a suitable formalism for modeling and querying such structures [29, 30]. In accordance with the Bunemans approach a semi-structured database is represented in a form of a rooted edge-labeled graph, like in Fig.1, and a schema is a rooted graph whose edges are labeled with formulas. A database SDB conforms to a schema SS if there is a correspondence between the edges in SDB and SS, such that whenever there is an edge labeled a in SDB, there is a corresponding edge labeled with predicate p in SS such that p(a) holds.
Figure 1: Three different kinds of semi-structured data: a- an example where the resource has two attributes; bsimplest graph; c-example of multiple instances of the same type.
c) a)
Link 1
Resource A Resource B Link 2 Resource C Link 4 Value 1 Link 3 Value 2 Value Value 1 Value 2 Value 3 Value 4 Link 22 Resource of type A

b)
Link 11

Resource A1 Resource C1

Resource of type B
Resource of type C Resource of type C

We can naturally set a correspondence between elements of the RDF semi-structured database SDB and the knowledge base KB: the set of concepts C and the set of instances I become the nodes of the graph, and the set of relations R maps to the graphs edges. In this context the general problem of achieving semantic interoperability can be looked at as the task of graph-based ontology transformation. The source graph of the semi-structured database comprises messages from the sensors networks, and corresponds to the ontology of IT-infrastructure with the knowledge base KB1 = = C1, R1, I1, C1, R1. Another ontology expresses the users domain of discourse in terms of the knowledge base KB2 = C2, R2, I2, C2, R2, and defines permitted structure of the destination graph. Applying the formal approach and visual cues of graph transformations [33, 34] we may define a transformation workflow process with the aim to produce the destination graph of semi-structured database expressed in terms of users ontology. In our opinion the transformation workflow process should consist of two different kinds of operations: the data transformation and the structure transformation. The data transformation uses leaf's values of the source graph, namely the literals of the RDF model, to produce values of the destination graph, namely literals of another RDF model. The data transformation is described as a sequence of interrelated data processing operations that can include elementary RDF literals transformation functions as well as access operations to external databases and Commercial Out-of-Shelf Software. The structure transformation uses location information of different subparts of the source graph to build the corresponding subparts of the destination graph.

3. Description of Methodology
Concerning implementation and design issues, we share the opinion [16,17], which says that software engineering knowledge representation, its transformation and automation of systems design can be achieved by application of formal model-driven approaches examining hierarchies of UML-based meta-models, models and ontologies. In our approach to consistent design and development of the Ontology Mediators family, we apply a few foundation principles for unification of work activities and formal methods. First of all, the consistent three-level UML modeling paradigm is used to create all information models and ontologies. At the top level M3 UML Meta-Object Facility provides for a single consistent meta-meta-model; at the middle level M2 UML profiles play a role of meta-models

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

480

and specify domain languages for various aspects descriptions; at he level M1 a collection of UML models represents formal description of IT-infrastructure, business view and architecture concepts of Ontology Mediator. The second distinctive feature of our methodology is coupled consideration of classes together with instances during modeling of IT-infrastructure and Enterprise business aspects. It allows for application of formal methods based on Information Flow theory [5]. Moreover, simultaneous working with classes and instances facilitates natural application of semi-structured data models. Representation of domain concepts and actual data in the context of the same semi-structured data model gives as a practical opportunity to employ a powerful technique of graph-oriented transformations in order to define rules of ontology transformation and methods of mapping between different models.
Figure 2: Proposed methodology of Ontology Mediators design and implementation. Dotted filling denotes initial core activities; the dark color defines repeatable activities during design of specific Ontology Mediator. UML models of the level M1 are the results of the activities of the correspondent track.
M3 M2 M1
Transformation profile Transformation models Track #1 Conceptualization of OM acrhitecture OM arch. profile Architecture models UML Meta-Object Facility Technologyspecific profiles Infrastructure models Track #2 Conceptualization of global ITInfrastructure Businessprocess profile Enterprise Ontology Track #3 Conceptualization of Enterprise Business aspects Conceptualization of users domain Classification of shared instances Matching Enterprise & users concepts Ontology UML profile Users Ontology

Classification of shared instances

Matching IT & Enterprise concepts

Core Activities
Selection of IT components for communication Design of a source semistructured model

Design of a destination semistructured model

Design of a semantic transformation model

Design of OM platform-specific models Semi-automated generation of artifacts Coding, assembling, deployment

481

Eduard Babkin and Habib Abdulrab

Fig.2. illustrates general principles of the proposed design methodology, in which three main tracks of modeling and design activities can be recognized. All three tracks begin from conceptualization of correspondent domains of discourse. Conceptualization activities of the first track produce UML models with architecture concepts of Ontology Mediator. The architecture models describe reusable software and hardware components and define extensibility interfaces for future use. The second track includes analysis of Enterprise business aspects and producing whole enterprise ontology. Based on Ontology UML profiles, that ontology describes structure and behavior of the appropriate knowledge domain, expressed in form of appropriate UML concepts (classes and logic constraints). The third track starts from conceptualization of underlying IT-infrastructure in terms of specialized UML models. Following our methodology the designer should perform Core Activities (conceptualization of IT-infrastructure, Enterprise business aspects and architecture of Ontology Mediator) only for the first time of methodologys application in new domain. But specification of UML-based users domain ontology is the repetitive activity, and it precedes development of any new Ontology Mediator device. In terms of the produced ontology shared instances are classified and matching between enterprise ontology and users ontology is performed. In the result the developers can select necessary elements of IT-infrastructure to communicate with, and they can design a correspondent source semi-structured data model in terms of IT-infrastructure models. By a similar way, the destination semi-structured data model is developed in terms of the users ontology. Once the source and destination models are produced all three tracks are joined, and mutual design of a semantic transformation model is performed. In terms of specialized UML profiles this model describes a workflow transforming the source model to the destination model. In fig.3 a fragment of the correspondent transformation profile is represented.
Figure 3: The Transformation UML Profile
OMTransform URIAssociation OMTransform Association * 1 * 1 OMTransform ExtractLexical

OMTransform Transformation

Atoms 1

Output
*

Input
OMTransform CompositionLexical OMTransform AggregateCompositionLexical

OMTransform GenericAtom

OMTransform StructuralAtom

OMTransform Copy

OMTransform Constant

OMTransform Numeric

OMTransform Logical

OMTransform Function

OMTransform EnvConst OMTransform RuntimeConst OMTransform AggregateNumeric OMTransform ComparatorLogical OMTransform GeneralFunc OMTransform GatewayFunc

As soon as the users approve the content of the semantic transformation model it becomes a basis for design of platform specific models, describing a concrete instantiation of Ontology Mediator. The platform-specific models determine which reusable hardware and software components will be combined together and which extension interfaces should be implemented during manual coding. Once the platform-dependent models have been produced the platform designers apply generation algorithms, which produce a skeleton of source code, as well as a list of additional hardware components and recommendations for selection of suitable base hardware modules. During final assembly of Ontology Mediator the programmers write small portions of glue code, extending the generated skeleton, and the hardware engineers equip base hardware modules with additional

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

482

components. This stage finishes process of Ontology Mediators development and the ready end-user product is shipped to the customer or is deployed in to the existing IT-infrastructure.

4. Framework Architecture
According to the proposed methodology we developed a consistent ontology-based framework, supporting design and development of Ontology Mediators. The framework has client-server architecture, and consists of front-end and back-end components, promoting team work (fig.4).
Figure 4: Ontology-based framework for design and development of Ontology Mediators

Standard Interface

Analyzer Server

Server Connection Plugin Workbench Workspace and Runtime Rational Software Architect The front-end

SETRAL Plugin

Reusable Transformation Engine

Platform Generator Server

OM Platform hardware modules

The back-end

The graphical front-end supports design of ontologies and UML-models for semantic transformations. Based on the Rational Software Architect platform by IBM and Eclipse Modeling Framework, the front-end adds several plugins, which implement a new visual modeling principle called Semantic Transformation Lasso (SETRAL), and provide interfaces to the framework back-end. SETRAL is extremely useful when modern Tablet PC and Visual Interactive Desks can be used during conceptualization and matching different ontologys concepts. For the user SETRAL offers a special smart graphical tool - Semantic Lasso (SL). That is a closed shape with arbitrary smooth boundaries. The user can draw a free-hand fragment of the SL boundary (the path) on top of the source UML class model to select certain classes to be matched. SETRAL automatically closes the path has been drawn and makes it smooth as necessary. Additionally SETRAL analyzes the source model and the metamodel to detect other entities (classes) that are semantically closely related to the entities inside the SL, but were not manually selected by the user. SETRAL proposes to include related entities with automatic extension of the SLs shape as needed. As soon as the source content of the SL was defined, SETRAL performs semantic mapping of the source entities inside the SL to the correspondent entities of the destination model. The result is displayed inside the SL as a fragment of the destination UML model. The level of transparency inside the SL can be adjusted interactively, so the user sees either two models or only one of them. Inside the SL the user can select entities of the destination model and switch to another editors to see the complete result of the matching. He/She has a possibility to easily return to the original editor with the active SL. SL tool can be used also for comprehensive analysis of the class instances. In this case after definition of the classes inside the SL, the user can run special SETRAL algorithms that will automatically generate instances with populated attributes. The framework back-end contains the Analyzer Server and the Platform generator Server. The former server is closely related with the design front-end. That component implements mathematical models of ontologies matching and models transformations as well as provides interfaces for loading core meta-models, domain models, domain-specific constraints in terms of OCL, as well as specifications of target software and the hardware platform. In accordance with loaded models the Analyzer Server automatically finds correspondent classes and attributes, returning results to the

483

Eduard Babkin and Habib Abdulrab

design GUI for further analysis by experts. The Platform Generator Server adopts basic principles of Model-Driven Architecture and automatically generates software artifacts for the selected target transformation platform. Each kind of the supported target platform uses the same library of software components (T-Engine). T-Engine contains algorithms for runtime semi-structured data transformation and business process enactment. Components of T-Engine can be deployed into the application server, the multi-agent system, or into the embedded platforms. In the later case our own specialized hardware platform is used.

5. T-Engine: architecture and implementation concepts


T-Engine is parted into four architectural layers, providing for different aspects of semantic interoperability: the External Communication Media layer, the RDF-mediation layer, the Internal Communication Media layer, and the Semantic Transformation layer (fig.5).
Figure 5: Detailed architecture of T- Engine
External Comm. Media Control Interface

Adapters
JMS MOMInterface Java Space CORBA OOInterface EJB JDBC ODBC DBMSInterface

RDF-Mediation

Proxies
WebService OSS Service Managing Agents CORBA Service

Internal Comm. Media

Event Interfaces
JMS Java Space

Data Interfaces Communication Management


JDBC WEBService

Semantic Transformation

Semantic Transformation Management

STE

Data Brokers

Containers (RDF, beans)

The External Communication Media layer gives the implementation-neutral interface to outer distributed components of UCE. This layer comprises three unified operational interfaces to deal with major classes of modern distributed technologies (i.e., Message-Oriented Middleware, remote calls of objects methods, Database Management Systems), as well as so-called adapters. Each adapter implements certain operational interface by means of concrete libraries and vendor-specific tools. A separate module implements control and management interface of T-Engine. The RDF-Mediation layer provides for encapsulation of heterogeneous data structures and communication algorithms in accordance with a single message-oriented paradigm. Following this paradigm the correspondence is set between every communication act with the external component and the message of particular type. Interacting with external components, the RDF-Mediation Layer creates one or many messages instances, and passes them to the Internal Communication media Layer for

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

484

further use. Similar to the adapters, at the RDF-Mediation Layer a special component called Proxy takes responsibility of messages production for a certain technology. The produced message consists of actual data embodied in the form of RDF, and internal metainformation (e.g. type qualification, messages timestamp or a messages producer). The type of the message poses restrictions on its possible data structure. These restrictions are expressed in the form of the RDF schemata and define allowable atomic fields and sub-structures for any valid messages instance of this type. Thus the messages type can be concerned likely a micro-ontology, and it can be created and modified in the framework front-end during designing the source semi-structured model. Generalization of information interchange in the form of messages flows promotes uniform representation of UCE components behavior and data inside the underlying layers: components play roles of message consumers or producers and their interaction can be uniformly represented as modification, creation or transformation of messages instances. The Internal Communication Media layer supports asynchronous message-based communication between the layers and provides for transaction management. Interacting with the Communication Management module, different modules of T-Engine subscribe to messages of particular type. Being notified of occurrence of new messages, the modules-subscribers fetch the messages from the internal queue, transform them, and put newly produced messages back for shared use. To improve performance and reduce the message-processing costs the Communication Management module also promotes separation of event and data flows as much as possible, avoiding passing large RDF models inside the message body at the intermediate stages of the message processing. Once submitted to the Communication Management module, the message is analyzed and rearranged to store either the original RDF model or merely short pragmatic instructions. In the later case the module of Data Interfaces uses pragmatic instructions for the lazy information retrieval from external data sources in order to reconstruct the complete RDF model. In the module of Event Interfaces all intermediate messages processing tasks such as routing, filtering, collecting are performed on the basis of the internal meta-information without touching upon the correspondent RDF model. Hence the creating of the content of the huge RDF model can be postponed up to the moment of its actual use, but in the case of the small RDF model its content is completely stored inside the message. The Semantic Transformation layer plays the central role in the successful fulfillment of Ontology Mediators activities. In particular, on this layer the Semantic Transformation Management module collects instances of input messages and fuses them into the united RDF model, which is in direct correspondence with the source semi-structured model, produced as the result of our design methodology. Relating messages instances with each other, the Semantic Transformation module exploits the messages meta-information and capabilities of RDF framework to link different resources via universal resource identifiers (URI). Once all needed instances have been collected and the source RDF model has been completely composed in accordance with the specification, the Semantic Transformation Management module places the model into the container for further processing by means of so-called Semantic Transformation Entities (STE). Each STE serves as an independent workflow process, reacting on completed models appearance in the particular container. T-Engine allows dynamic independent deployment of different STEs and later manages their concurrent execution. STEs can be organized into chains of linked semantic transformations, establishing complex information processing inside Ontology Mediator. Through the interface of DataBrokers a single STE gains access to the elements of the united RDF model, and performs transformation of this model to the destination RDF model in accordance with the semantic transformation UML model. When the STE successfully finishes building of the destination RDF model, algorithms of the Semantic Transformation Management module split this model into separate RDF sub-models, corresponding to distinct messages. Carrying on the information in terms of the users ontology, messages proceed to the External Communication Media layer. Alternatively, the messages can take part in construction another source RDF model.

485

Eduard Babkin and Habib Abdulrab

The architecture of T-Engine, as it has been described in previous paragraphs, is almost technology independent. Indeed, different modern distributed technologies such as CORBA, EJB, JMS and JINI can be used for practical implementation. In the course of various software prototypes design we have found that although CORBA and EJB are more widely used, the JINI-based implementation gives many attractive features. We have discovered that JINI technology is the most suitable for implementation of proxies on the RDF-Mediation layer. In this case JINI Smart Proxies can be installed quickly for preparation of the RDF-model. Smart JINI proxies expose a unified interface to Event Interfaces, Control Interfaces and Data Interfaces. These proxies are available via JINI lookup service (reggie). CORBA and JMS implementations of Event Interfaces were also performed but they are not described here.

6. The Extensible Hardware Platform for Ontology Mediator


Our investigations of several use cases determined major guidelines for hardware architecture of Ontology Mediators. It should have nearly the same customer properties as sensor network devices: reasonably low prices, friendly interface for installation and maintenance. These propertyies allow for preserving unique features of sensors networks and deploy Ontology Mediators in different locations, following changing and diverse needs of the end-users. At the same time Ontology Mediator should provide for a wide range of different mediation scenarios, and, in our vision, its hardware should be principally able to support communication via any of the most popular wireless protocols as well as wired ones (i.e. Ethernet, RS232/435, CAN). Later requirement leads to broad variations in internal hardware and software architecture. Fitting the requirements above and the proposed model-driven design methodology as well, the extensible hardware platform comprises two specially designed hardware modules: Processing Unit an unmodified part of Ontology Mediator, where the central processor and basic communication interfaces are installed. Extension Board a customizable interface board that can be easily extended by different hardware and embedded software components (plug-ins) on demand of specific requirements. A number of design solutions were studied and prototypes were developed in order to find the most suitable decomposition of functions and balanced costs. Our latest experimental implementation of Ontology Mediators Processing Unit has the following features (Fig.6-a): MCU: LPC2294 16/32 bit ARM7TDMI-St with 256K Bytes Program Flash, 16K Bytes RAM, 4x 10 bit ADC, 2x UARTs, 4x CAN, I2C, SPI, up to 60MHz operation. 4MB SRAM 4x K6X8008T2B-F/Q SAMSUNG. 4MB TE28F320C3BD90 C3 INTEL FLASH. 10Mb TP Ethernet (CS8900A). OS: RTOS ECOS 2.0. Connectors: Power supply, RJ45, HDR26F (for connection to the extension board). For the developed Processing Unit a reduced version of T- Engine was developed. In that version transformation algorithms, internal control and interfacing functions were programmed in C and C++. Now we are at the final stage of porting Wonka Java virtual machine [21] to the Processing Unit that will allow enriching application capabilities and provide full-fledged implementation of TEngine.

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

486

Figure 6: Hardware components of the extensible platform: a- Processing Unit in the case of stand-alone usage without Extension Board; b an experimental sensor.

a)

b)

During experiments with Ontology Mediators we paid attention to semantic transformation algorithms, and used multiple wired sensors to reduce implementation costs. So, now Extension Board has support of both wired and wireless protocols: I2C, two CAN, 8x30 sensors (Fig.6-b), SPI-to-UART MAX3000 transceiver for connection to wireless SANET via radio transceiver. We place to our nearest plans generalization of the board architecture and its redesigning in the framework of evolvable hardware paradigm. The final version of Extension Board will include FPGA-based evolvable hardware part and a set of free slots for drivers and transceivers. Most of the experiments with ontology mediation algorithms employed two wired SANETs, directly connected to Ontology Mediator via RS485 interface. The first test-bed SANET with ring topology consists of 70 sensors and actuators deployed to a rail road model. Ontology Mediator presents dynamic state of the rail road in terms of an operator and allows performing basic actions by friendly Web-based interface (Java applets). The second SANET has star topology with eight rays. Each ray is capable of supporting up to thirty sensors. This network was installed on a real test site to monitor environmental conditions. In that case, Ontology Mediator (Fig.7) provides operators with real-time information about operational conditions, as well as equipment performance degradation forecasts. For these specific purposes, specialization of Ontology Mediator includes equipping Extension Board with LCD monitor and controls buttons, developing models and ontology mapping between low-level measurements of temperature, humidity and gas contamination to high-level abstractions of equipment operators to integrate SANET with external information systems.

487

Eduard Babkin and Habib Abdulrab


Figure 7: Specialization of Ontology Mediator for telecommunication domains.

7. Conclusion
In this work we focused the attention on the problem of achieving semantic interoperability in the context of wide-area Ubiquitous Computing Environment. The concept of Ontology Mediator was offered to designate a specialized device or a software component, which goal is fusion of sensors data, their smart transformation and expression in various forms of real-world concepts by application of ontologies manipulation. The developed design methodology and supporting framework provide foundations for model-driven development of Ontology Mediators. The basis of our approach is the ontology transformation process with locally defined conditions for joining separate input messages into the united RDF model and explicitly defined rules of its transformation into the set of output RDF models. Implementation of that process was done in the library of reusable software components (TEngine). The T-Engines components implement original message-oriented algorithms for semantic consistency testing, consolidation and transformation. During software implementation of T-Engine different distributed technologies were exploited and the tuple space based communication paradigm based on JavaSpaces technology showed the best performance and was the most attractive one from the architectural point of view. To study effects and benefits of embedded T-Engines implementations we developed an extensible specialized hardware platform. In comparison with known middleware architecture approaches for sensors networks [6, 13] our solution operates on a level of meta-data and allows bridging a gap between different information object-oriented models. In fact, having interfaces to different communication protocols, Ontology Mediator can be applied for orchestrating heterogeneous middleware services when the environment consists of different sensors networks and enterprise-wide applications. Increased reusability is seemed to be another attractive feature of our approach. During the Ontology Mediators design, the platform designer accumulates knowledge about certain application domains and patterns of behavior for different customers and needs. It allows continuous extending library of available models; moreover third party developers and teams can share such library. At the same time application of the single meta-model for development of different models simplifies models transformation and integration. Despite of using the same general ontology paradigm for meta data representation proposed in [18], we contribute original design methodology and the software-hardware framework that allow implementing solutions of various scales, performing ontology transformation simulations and testing. The generic layered approach to design allows easy adaptation of the framework for different distributed systems. Regardless of components structures modification and evolutionary variations in semantics, the developed solution allows for maintaining the permanent information consistence among multiple components. For example, in the case of modification of the underlying database

Ontology-Based Framework for Integration in Ubiquitous Computing Environment

488

schema or the interface definition, it is necessary only to modify the definition of semantic transformation for some STE without touching upon the implementation of external components. Solutions like TSpaces (IBM), MARS-X[31] and XMIDDLE[32] properly illustrate trends in application of modern Internet technologies, such as XML and RDF, in coordination frameworks. In our opinion MARS-X has some common points of interest and can be compared with our framework. Both approaches use the tuple-based coordination paradigm based on JavaSpaces software implementation. In MARS-X tuples are used for the XML document storing, and different software agents can extract relevant information, using complex search patterns defined programmatically. In our approach tuples are used to notify about the message instance presence and components define conditions for joining messages on the basis of its properties by a declarative way. As for semantic interoperability the important restriction of MARS-X is necessity to use the same structure of XML document (share the same DTD) for all components of a distributed system. Modification of DTD will require redesign of software components that is not always possible in the loosely coupled system. In our case the declarative description of joining conditions and delegating messages collecting and transformation activities into the separate middleware service allows to separate design of component algorithms from management of semantic interoperability. Considering implementation issues of our framework, it can be mentioned that the designed software architecture permits rapid inclusion of new software technologies on different levels without significant changes of the core. Splitting the process of the message processing from the process of huge RDF models retrieving makes Ontology Mediator more robust.

8. Acknowledgement
This work was partly supported by the research grant #07-07-00058 of Russian Foundation of Basic Research.

489

Eduard Babkin and Habib Abdulrab G.D. Abowd, E.D. Mynatt, Charting Past, Present, and Future Research in Ubiquitous Computing, ACM Transaction on Computer-Human Interaction, vol. 7, No 1, pp.29-58, 2000. E.Niemel, J.Latvakoski, Survey of requirements and solutions for ubiquitous software, in Proc. of the 3rd international conference on Mobile and ubiquitous multimedia, College Park, Maryland, 2004, pp. 71-78. T. Jeng, Designing a Ubiquitous Smart Space of the Future: The Principle of Mapping, in Proc. of International Conference of Cognition and Computation (DCC 04), MIT, Cambridge, 19-21 July, 2004. Y. Kawahara, M. Minami, S. Saruwatari, H. Morikawa, and T. Aoyama, Challenges and Lessons Learned in Building a Practical Smart Space, in Procs. of the 1st Annual International Conference on Mobile and Ubiquitous Systems:Networking and Services (MobiQuitous 04), Boston, Massachusetts, August 22-25, 2004 , pp. 213-222. J. Barwise, J. Seligman, Information Flow, Cambridge University Press, 1997. J. W. Branch, J. Davis, D. Sow, and C. Bisdikian, "Sentire: A Framework for Building Middleware for Sensor and Actuator Networks," in Proc. of the 1st IEEE International Workshop on Sensor Networks and Systems for Pervasive Computing (PerSeNS'05), Kauai Island, Hawaii, March 8-12, 2005, pp. 396-400. Misc.Tinyos: A component-based os for the networked sensor regime. In http://webs.cs.berkeley.edu/tos/. J. Feng, F. Koushanfar, and M. Potkonjak, System-Architectures for Sensor Networks Issues, Alternatives, and Directions, in Proc. of the IEEE International Conference on Computer Design: VLSI in Computers and Processors (ICCD02), Rio de Janeiro, Brazil, September 1618, 2002, pp. 226. S. Tilak, N. Abu-Ghazaleh, W. Heinzelman,A Taxonomy of Wireless Micro-Sensor Network Models, ACM Mobile Computing and Communications Review, vol. 6, No.2, 2002, pp. 2836. Crossbow Technology,Inc., In http://www.xbow. com/Products/products.htm. C. Curino, M. Giani, M. Giorgetta, A. Giusti, A.L. Murphy, and G.P. Picco,Tiny LIME : Bridging Mobile and Sensor Networks through Middleware, in Proc. of the 3rd IEEE International Conference on Pervasive Computing and Communications (PerCom 2005), Kauai Island, Hawaii, March 8-12, 2005. pp. 61-72. V. Tsetsos, G. Alyfantis, T. Hasiotis, O. Sekkas, and S. Hadjiefthymiades, Commercial Wireless Sensor Networks: Technical and Business Issues, in Proc. of the Second Annual Conference on Wireless On-demand Network Systems and Services (WONS05), St. Moritz, Switzerland, January 2005, pp. 166-173. S.I. Ahamed, A. Vyas, and M. Zulkernine, Towards Developing Sensor Networks Monitoring as a Middleware Service, in Proc. of the 2004 International Conference on Parallel Processing Workshops (ICPPW 04), Montral, Qubec, Canada, August 15-18, 2004, pp. 465-471. E. Tokunaga, A. van der Zee, M. Kurahashi, M. Nemoto, and T. Nakajima, A Middleware Infrastracture for Building Mixed Reality Applications in Ubiquitous Computing Environments, in Procs. of the 1st Annual International Conference on Mobile and Ubiquitous Systems:Networking and Services (MobiQuitous 04), Boston, Massachusetts, August 22-25, 2004 , pp. 382-391. E. Chan, J. Bresler, J. Al-Muhtadi, and R. Campbell, Gaia Microserver:An Extendable Mobile Middleware Platform, in Proc. of the 3rd IEEE International Conference on Pervasive Computing and Communications (PerCom 2005), Kauai Island, Hawaii, March 8-12, 2005. pp. 309-313. I. Oliver, Applying UML and MDA to Real Systems Design, in Proc. of the conference on Design, Automation and Test in Europe (DATE'05), Munich, Germany, 7-11 March, 2005, vol. 1, pp.70-71.

References
[1] [2] [3] [4]

[5] [6]

[7] [8]

[9] [10] [11]

[12]

[13] [14]

[15]

[16]

Ontology-Based Framework for Integration in Ubiquitous Computing Environment [17] [18]

490

[19] [20]

[21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]

P. Volgyesi, A. Ledeczi, Component-Based Development of Networked Embedded Applications,in Proc. of the 28th Euromicro Conference (EUROMICRO02), Dortmund, Germany, 4-6 September, 2002, pp. 68. H. Chen, F. Perich, T. Finin, and A. Joshi, SOUPA: Standard Ontology for Ubiquitous and Pervasive Applications, in Procs. of the 1st Annual International Conference on Mobile and Ubiquitous Systems:Networking and Services (MobiQuitous 04), Boston, Massachusetts, August 22-25, 2004, pp. 258-267. Y. Zhou, Q. Zhao, and M. Perry. Reasoning over Ontologies of On Demand Service, in Proc. of 2005 IEEE International Conference on e-Technology, e-Commerce and e-Service (EEE'05), Hong Kong, 29 March-1 April, 2005, pp. 381-384. N. Hnle, U.-P. Kppeler, and D. Nicklas, Thomas Schwarz, Matthias Grossmann,Benefits of Integrating Meta Data into a Context Model, in Proc. of the 3rd IEEE International Conference on Pervasive Computing and Communications (PerCom 2005), Kauai Island, Hawaii, March 812, 2005, pp. 25-29. Wonka Java VM Project. In https://opensource.luminis.net/confluence/display/WONKA/Home C. Herring, "Microprocessors, Microcontrollers, and Systems in the New Millennium," IEEE Micro, vol. 20, no. 6, 2000, pp. 45-51. J. F. Sowa, Knowledge Representation: Logical, Philosophical, and Computational Foundations, Brooks Cole, Pacific Grove, CA, 2000. Y. Kalfoglou, Exploring ontologies, in Handbook of Software Engineering and Knowledge Engineering, Vol. 1, Fundamentals, ed. S.K. Chang, World Scientific, Singapore, 2001, pp. 863887. T. R. Gruber, Toward Principles for the Design of Ontologies Used for Knowledge Sharing, International Journal of Human-Computer Studies, 43(5/6), 1995, pp. 907-928. G. Dragan, D. Dragan, D. Vladan, Model Driven Architecture and Ontology Development., Springer-Verlag. (ISBN: 978-3-540-32180-4), 2006. M. Ehrig, Ontology Alignment : Bridging the Semantic Gap. Series: Semantic Web and Beyond, Vol. 4. Springer-Verlag. (ISBN 978-0-387-32805-8), 2007. S. Abiteboul, R. Hull, V. and Vianu, Foundations of databases., Addison Wesley Publ. Co., Reading, Massachussetts, 1995. P. Buneman, S. Davidson, M. Fernandez, D. Suciu, Adding structure to unstructured data., Technical Report MS-CIS-96-21, University of Pennsylvania, Computer and Information Science Department, 1996. P. Buneman, Semistructured data, Proceedings of the ACM Symposium on Principles of Database Systems, 1997. G. Cabri, L. Leonardi, F. Zambonelli, XML dataspaces for mobile agent coordination., In Procs. of the 2000 ACM symposium on Applied computing, 2000, pp.181-188. C. Mascolo C., L. Capra, W. Emmerich, An XML-based Middleware for Peer-to-Peer computing., In Procs. of the First Int. Conf. on Peer-to-Peer Computing (P2P'01), 2001, pp. 69-74. G. Rozenberg G. (ed.),Handbook of Graph Grammars and Computing by Graph Transformations, Volume 1: Foundations. World Scientic, 1997. B. Hoffmann, M. Minas, Towards Generic Rule-Based Visual Programming, 2000 IEEE International Symposium on Visual Languages (VL'00), 2000, pp. 65-67. K.S. Candan, H. Liu, R. Suvarna, Resource description framework: metadata and its applications, ACM SIGKDD Explorations Newsletter, 3(1), 2001, pp. 6-19.

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.491-508 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams
V.O.S. Olunloyo Faculty of Engineering, University of Lagos Department of Systems Engineering O.Damisa Faculty of Engineering, University of Lagos Department of Mechanical Engineering C.A. Osheku Faculty of Engineering, University of Lagos Department of Mechanical Engineering A.A. Oyediran AYT Research Corp., McLean VA. 22102, USA Abstract The earlier work on static analysis of slip damping with clamped laminated beams is herein completed by considering the case where the pressure at the interface is not uniform. The importance of this work derives from the fact that experiments have confirmed that interface pressure is hardly constant, but infact has a non-uniform distribution that is better simulated by some of the profiles discussed in this paper. Furthermore, it is shown that when the interface pressure is either a polynomial or a combination of hyperbolic functions, it is possible to obtain higher energy dissipation and lower cantilever deflection than would obtain for constant interface pressure profile. Keywords: laminated beams, non-uniform interface pressure, slip, energy dissipation.

1.0. Introduction
In our earlier work, the static analysis of clamped laminated beams was presented as a model for investigating noise dissipation in turbine blades for supersonic and advanced subsonic engines. Renewed interest in the problem derives from the recent work of Damisa [1] and the interest in treating the model in a more general context of energy dissipation by interfacial friction in laminated structures. Most of the early work, including that of Goodman and Klumpp [2], was for the case of uniform interfacial pressure between the laminates. One way of simulating such a pressure distribution is through the introduction of bolted joints at appropriate locations along the laminate interface. However, there has been experimental confirmation that interface pressure as met in practice is rarely uniform. Consequently, there is some justification in examining the effect of non-uniform pressure distribution at the interface for the model problem under investigation.

492

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran

In fact, recent studies by Shin et al [3] and later Nanda and Behera [4] reported on the pattern and intensity of pressure distribution at the interface of bolted joints. In one case, theoretical expression for the pressure distribution at the interface using numerical data of Ziada and Abd [5] was reported by Nanda and Behera, which confirmed that the pressure field around a bolted joint is not uniform. However, by spacing the location of such joints in a contrived manner, it is still possible to arrange a superposition of such pressure fields to achieve an overall uniform pressure distribution. Following this approach, Nanda and Behera [4] used the results of Ziada and Abd to calculate the required spacings between their bolted joints that would allow them to correctly simulate a uniform pressure field at the interface. The damping capacity of such layered and jointed structures was thereafter evaluated both numerically and experimentally using various parameters. This then means that if the engineer or designer is to retain his freedom to locate such joints wherever he wishes, he must be prepared to confront the reality of non-uniform pressure distribution along the laminate interface. In the present work, we limit our analysis to two interface pressure distribution profiles and examine their effect on the response, relative slip and energy dissipation emanating from the slip damping under static conditions namely: (a ) P ( x ) = P0 (sinh 1 x + B cosh 2 x ) (1)
( b) P( x ) = P0 am x m
m=0 M

(2)

In choosing these distributions it is assumed that the pressure profile is positive definite over the entire length of the laminate interface i.e, P(x ) > 0 ; 0 x L

2.0. Case of Interface Pressure Distribution

P( x ) = P0 (sinh 1 x + B cosh 2 x ) Fig. 1 shows the geometry and coordinate axes for a layered beam of width b, whilst h is the uniform depth of each layer. The layers are held together by a pressure distribution P(x) applied to both the upper and lower surfaces as illustrated below:
Figure 1: Laminated Beam of Similar Material, Coordinate Axes and Geometry

2.1. Analysis of Response Under Static Conditions

For the case illustrated above, we derived in our earlier paper [6], the corresponding governing differential equation which for this case can be written as:

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams d 4W dx 4 = dP dx = P0 ( 1 cosh 1 x + 2 B sinh 2 x )

493

(3)

where

6 Eh 2 We next introduce the Fourier Finite Sine Transform namely: L 2 nx nx s {[]} = [] = []sin (4) dx , [] = []sin L L L n =1 0 so that on invoking the usual cantilever boundary conditions, and applying the finite sine transform on equation (3) we obtain: n n 3 3 n 4 4 F n +1 W xx (0 ) W ( n ) 3 ( 1) W (L ) + 4 L L L = P0 s {1 cosh 1 x + 2 B sinh 2 x} (5)

= P0 (1 + 2 )

where
1

1 ( 1)n cosh 1 L = 1nL 2 L2 + n 2 2 1

(6)

and

( 1)n sinh 2 L (7) 2 = 2 BnL 2 2 L + n 2 2 2 The analysis of equation (5) is further simplified if we apply the Goodman and Klumpp end condition namely: h F ( xz ) dz = 2b at x = L 0

(8)

and recall the shear stress relation namely: 1 P( x ) ( xz )1 = E (z 2 hz )W xxx (z h ) 2 h (9) Next, we substitute for the pressure profile defined in equation (1) into the expression for the shear stress so as to rewrite equation (8) as: h h 1 F E z 2 hz W xxx dz P0 (sinh 1 x + B cosh 2 x )( z h )dz = at x = L 2 h 2b 0 0

(10)

which can be further integrated to give L h L h 1 E (z 2 hz )W xxx dzdx P0 (sinh 1 x + B cosh 2 x )( z h )dzdx 2 h 0 0 0 0
F = dx 2b 0
L

(11)

Thus we find 6 Pav 6F W xx (0) = L 3 Eh 2 Ebh where

(12)

494

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran

Pav

P0

1 2 L

(1 B sinh 2 L 2 (1 cosh 1 L ))

(12a)

is a measure of the average pressure across the length of the interface. If we substitute appropriately for Wxx(0) from equation (12) above into equation (5), its Fourier inversion then gives ( 1)n+1 sin nx 2W (L ) n L n =1 nx sin 6 Pav 6F L 2 L3 W (x ) = 3 Eh 2 n =1 n 3 3 Ebh nx 3 sin 12 L + n 4 4L Eh 2 n =1 where

(13)

1 1 ( 1)n cosh 1 L 2 B( 1)n sinh 2 L 1 + 2 = = L n 12 L2 + n 2 2 22 L2 + n 2 2 (13a) To further simplify the series in equation (13) we next invoke the well-known closed form Fourier Series representations namely: n +1 1 ( 1) x = sin nx , 0 < x <1 n =1 n (14) and sin nx 2 x x 2 x 3 = + , 0 x2 (15) n3 6 4 12 n =1 We now introduce the decomposition = Pav n where 1 1 ( 1)n cosh 1 L 2 B( 1)n sinh 2 L L 12 L2 + n 2 2 22 L2 + n 2 2 n = = P0 Pav [1 B sinh 2 L 2 (1 cosh 1 L )] 1 2 L

2 L2 + n 2 2 1 ( 1)n cosh 1 L 2 B 12 L2 + n 2 2 ( 1)n sinh 2 L = 1 2 L2 1 2 P0 1 L2 + n 2 2 2 L2 + n 2 2 [ 1 B sinh 2 L 2 (1 cosh 1 L )] We can also write equation (13) in the form: 6 Pav x x 2 x 3 6F 2W (L )x 2 L3 + 3 Eh 2 6 4 12 Ebh (16) W (x ) = n sin nx 12 L3 Pav + Eh 2 n 4 4 n =1

)(

)(

where
x= x L

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams By imposing the boundary condition

495

dW (0 ) = 0 in equation (16) we now evaluate the dx deflection at the end of the laminated cantilever beam as F P 6 av W (L ) = L3 Pav 3 n 3 3 2 2 Eh Eh n =1 n Ebh (17) Thus, we can now express equation (16) in the form: 3 F P x 12L3 L Pav 3 n 3 av 3x 2 x 3 3 2 2 Eh Eh Ebh n =1 n W (x ) = 3 n sin nx 12L Pav + Eh 2 n 4 4 n =1 (18a) or as x sin nx W ( x ) = (1 Pav ) 3x 2 x 3 12Pav 3 n 3 + 12Pav n 4 4 n n =1 n =1 n

(18b)

where the following non-dimensionalized parameters have been introduced viz:


W = W ( x )Ebh 3 ; Pav L3 F =

( )

Pav ; F bh

( )

n F bh

Thus, in the limit of uniform interface pressure, i.e. 0 the static response correctly assumes the form: W (x ) = {(1 Pav )(3 x 2 x 3 )} (19) as was presented in our earlier paper.
2.2. Slip under Static Conditions

The relative slip at the interface of the laminated beams is given by u ( x ) = u 2 ( x,0 ) u1 ( x,0 + )
Following Goodman and Klumpp, equation (20) can also be written as u ( x ) =

(20)

{( ) ( ,0 ) ( ) ( ,0 + )}d
x 1 x 2

(21)

By invoking the bending stress relations namely: 2 ( x )1 (x, z ) = E (2 z h ) W + Pav (x L ) (22) 2 h x 2 and 2 ( x )2 (x, z ) = E (2 z + h ) W Pav (x L ) 2 h x 2 we can now rewrite equation (21) in the form x d 2W ( ) 2Pav ( 1)d u ( x ) = h + 2 Eh d 0 Equation (24) then simplifies to the form: cos nx u ( x ) = (4Pav 3) x 2 2 x 12Pav 3 n 3 + 12Pav n 3 3 n n =1 n n =1

(23)

(24)

(25)

496 2.3. Energy Dissipation

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran

The energy dissipated per cycle, following Goodman and Klumpp admits the form

D = 4b P( x )u ( x )dx
0

(26)
= DEb 2 h 3 , then L3 F 2

If we introduce the dimensionless energy dissipation parameter, D equation (26) can be written as
D = 4 Pav u dx
0 1 1 2 Pav (4 Pav 3) x 2 2 x dx 12 Pav 3 n 3 dx n =1 n 0 0 = 4 1 n 2 +12 Pav 3 3 cos nx dx n =1 n 0 1

4 2 2 6 Pav 8Pav 36 Pav 3 n 3 3 n =1 n

(27)

2.4. Critical Clamping Pressure for No Slip

Since the energy dissipated is a product of the slip and interface pressure it follows that in the presence of interface pressure and friction we expect to have energy dissipation provided there is slip. Thus under these conditions, the energy dissipated can only be zero when there is no slip. The condition for no slip can then be inferred by setting equation (27) to zero This gives n Pav 6 8Pav 36Pav = 0 n 3 3 which has the non trivial solution

Pav

9 = 0.75 1 + 3 n 3 2 n
1

2.5. Analysis of Optimum Clamping Pressure under Static Conditions

The optimum clamping pressure can be found from the partial derivative of the energy dissipated if we set dD d 4 (6Pav 8Pav2 ) 36Pav2 n 3 = 0 (28) = 3 dPav dPav 3 n =1 n

Thus, we derive the general expression for the optimum clamping pressure as
9 (29) Pav = = 0.375 1 1 + 3 n 3 2 n so that in the limit of uniform pressure, .i.e. 0 , we recover the optimum clamping pressure 3 Pav = P0 = as was earlier reported by Goodman and Klumpp. 8
1

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams

497

Furthermore, relation (29) also indicates that for our results to be meaningful, the interface 2 pressure must be positive definite; thus the requirement 3 n 3 > must hold. 9 n =1 n 2.6. Maximum Energy Dissipation from Slip Friction

The maximum energy dissipation due to slip damping occurs at the optimum clamping pressure. To evaluate this, we substitute the value of the optimal clamping pressure as given by equation (29) into the expression for energy dissipation (i.e. equation 27) to obtain: n 4 2 2 Dmax = (6 Pav 8Pav ) 36 Pav 3 3 3 n =1 n
= 3 H2 2
1 1

(30)

where
H 2 = 1 + 9 3 n 3 = 1 + 9 = 3 n3 ; 2 n 2 n =1 n The expression shows that the maximum energy dissipated is modulated by only one parameter viz: . The parameter , as we have indicated earlier, is a function of n which derives from the pressure gradient along the length of the laminate interface as can be deduced from equations (3), (5) and (13a). In fact, in the absence of this pressure gradient, the maximum energy dissipated is independent of Coulomb friction coefficient and we can then revert to the result of Goodman and Klumpp. However, once the interface pressure is not uniform, it allows to modulate the energy dissipation as seen from the result (30) above.

3.0. Case of Interface Pressure Distribution


P( x ) = P0 a m x m
m =0 M M

We now focus attention on the power series pressure distribution of the form
P( x ) = P0 a m x m
m =0

(31)

which is the traditional representation invoked for curve fitting of experimental readings such as was reported by Nanda and Behera [4].
3.1. Analysis of Response under Static Response

For this case, the corresponding equation takes the form M 1 d 4W dP = = P0 ma m x m1 (32) dx dx 4 m =0 By recalling equation (4) we can apply the Fourier Finite Sine Transform on equation (32) to obtain: n 4 4 F n 3 3 n M 1 n +1 W ( n ) 3 ( 1) W (L ) + W xx (0) = P0 s ma m x m 1 4 L L L m =1 i.e.

498

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


M 1 n 4 4 F n 3 3 n n +1 W ( n ) 3 ( 1) W (L ) + Wxx (0) = P0 a m m L L4 L m =1

(33)

where
m 1 m 1 Lm m = k! ( 1)n cos(n + k ) (34) k +1 k 2 k =0 (n ) By following the same process outlined in the section 2, we arrive at the following solution for this case viz: ( 1)n+1 sin nx 2W (L ) n L n =1 nx sin 6 Pav 6F 2 L3 W (x ) = 3 L 3 2 3 Eh n =1 n Ebh 3 12 L sin nx + n 4 4 Eh 2 n =1 (35) where M P0 L M a m Lm m Pav = a m x dx = P0 L m=0 m=0 m + 1 0 and

M 1 m =1

ma

m x 3 12 Pav

In fact, we can further show that


W (x ) =

(1 P )(3x
av

x sin nx + 12 Pav 4 4 3 3 n n =1 n n =1

(36)

where x and Pav have the same dimensionalisation as introduced earlier and = F bh

( )

3.2. Evaluation of Slip and Energy Dissipation under Static Conditions

The relative slip at the interface of the laminated beams can be evaluated in the usual way as cos nx (37) u (x ) = (4Pav 3)(x 2 2 x ) 12Pav 3 3 + 12Pav 3 3 n n =1 n n =1 On the other hand, energy dissipated per cycle, following Goodman and Klumpp admits the form D = 4b P( x )u ( x )dx
0
L

(38)

If we recall the dimensionless energy dissipation parameter, introduced earlier in section 2.3, then equation (38) can be written as D = 4 Pav u dx
0 1

(39)

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams

499

so that substitution for u from eqn (37) then gives the expression for dimensionless energy dissipation as: 4 2 2 (40) D = (6 Pav 8Pav ) 12Pav 3 3 3 n =1 n From this result, we can also deduce that the condition for no slip is governed by the relation
9 Pav = 0.75 1 + 2
1 1

= 0.75 1H 3

where
=
n =1

n 3
3 1

and

H 3 = 1 + 9
2

3.4. Analysis of Optimum Clamping Pressure under Static Conditions The optimum clamping pressure can be found from the partial derivative of the energy dissipated if we set dD d 4 2 2 = (41) 6 Pav 8Pav 12Pav 3 3 = 0 dPav dPav 3 n =1 n Thus, we derive the general expression for the optimum clamping pressure as

9 Pav = 0.375 1 + (42) 2 from which Goodman and Klumpps result can be correctly recovered for the case of uniform pressure. Similarly, we can show that for this case, the maximum dissipation energy from slip is given by 3 Dmax = H3 2
1

4.0. Discussion of Results


The results of our analysis indicate that there are critical values of the normalized interface pressure Pav at which displacement, slip or dissipated energy disappear even when there is coulomb friction. These are saddle points that should be avoided if the objective is to dissipate energy. On the other hand there are optimal values for which we have maximum dissipation of energy through slip damping. A summary of the set of values for both categories is tabulated below but it is pertinent to note here that in compiling this table, we have recalled some of our earlier results derived in [6] as they provide an adequate background for comparative analysis.

500
Objective Normalised Interface Pressure P = P0 Uniform Pressure Pav = P0 P = P0 (1+ x )
Pav = P0 1 + 2 P( x ) = P0 (sinh 1 x + B cosh 2 x )
Pav = P0 Pav =

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


No slip (u = 0 ) Max energy Dissipation (D = Dmax )

P0 = 0.75 1 P0 = 0.75 1 H 1

P0 = 0.375 1 P0 = 0.375 1H 1

Pav = 0.75 1H 2

Pav = 0.375 1 H 2

(sinh x + B cosh x )dx


1 2 0

P0

1 2

[ 2 (cosh 1 1) + B1 sinh 2 ]
M

P( x ) = P0 am x m
m=0

Pav = 0.75 1H 3
M

Pav = 0.375 1 H 3

Pav = P0 am x mdx = P0
0 m =0

1 M

m =0

am m +1

Here,

H 1 = 1 + 261 ; H 2 = 1 + 9 ; H 3 = 1 + 9 320 2 2 In plotting the various curves for both the hyperbolic and power series pressure distribution profiles, we have made use of a parameter av for estimating the average gradient of the pressure distribution across the interface. In particular, for the hyperbolic pressure profile, we have defined P P av = 0 [ 2 cosh 1 + B1 sinh 2 ] 0
1 2 1
as this parameter which is found to be a major factor that influences the mechanism of slip damping. Figures 1 and 2 for example, display the deflection and slip for the hyperbolic profiles defined by the indicated choice of parameters. In Figure 3 the computed energy dissipation for the sample problem is displayed. The point to be made here is that although the largest energy dissipation occurred in the case of the most negative pressure gradient, the converse was the case in respect of deflection and slip. The implication of this is that by making an appropriate choice of values for the parameters defining the hyperbolic pressure profile, it is possible to simulate higher energy dissipation while at the same time obtaining lower deflection and slip than what obtains for the case of uniform interfacial pressure profile. This fortuitous coincidence is not always the case as can be seen from figures 4, 5 and 6 where the deflection, slip and energy dissipated are displayed for a power series interfacial pressure profile.

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams
Figure 1: Normalized static response profile for hyperbolic pressure form

501

Figure 2: Normalized slip profile for hyperbolic pressure form

502

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


Figure 3: Normalized energy dissipation for hyperbolic interface pressure

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams
Figure 4: Normalized deflection at optimum clamping pressure for power series interface pressure

503

Figure 5: Normalized slip at optimum clamping pressure for power series interface pressure

504

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran

Figure 6: Normalized energy dissipation versus normalized clamping pressure for power series interface pressure

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams

505

One mechanism for simulating such a negative pressure gradient where the beams are held together by bolts as in the case of Ziada and Abd, is to progressively reduce the torque on the securing bolts as one moves away from the clamped to the free end of the cantilever beam. One of the findings in our earlier paper was that the interfacial pressure gradient can play a significant role in the effectiveness of the slip damping. In fact, in Figure 7, a plot is given relating the scaling factors H 2 and H 3 to the pressure gradient. Furthermore, we find that the maximum energy that can be dissipated in each case a mere scalar multiple of the value 3 derived for uniform 2 interfacial pressure viz: 3 H 2 and 3 H 3 respectively for the hyperbolic and power series pressure profiles. Thus it is by analysing H 2 and H 3 that we can deduce how best to optimise the slip energy dissipation. This indicates that both and , which derive from the pressure gradient should have negative values for this to happen. However, the actual value is limited by the requirement that the pressure must remain positive definite across the entire length of the beam interface. Also, note from Figures 7a and 7b that in both cases, we correctly recover the value H 2 = H 3 = 1 at the point where the pressure gradient is zero; corresponding to the case of uniform interface pressure.

506

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


Figure 7a: Maximum energy dissipation scale factor for hyperbolic pressure gradient

Figure 7b: Maximum energy dissipation scale factor for power series pressure gradient

Further Results Concerning the Static Analysis of Slip Damping with Clamped Laminated Beams

507

References
[1] [2] [3] [4] [5] [6] Damisa, O.: Dynamic Analysis of Slip Damping with Reference to Elastic Beams, J. Engrg. Research (JER), 10, nos. 1 & 2, 1 14 (2002). Goodman, L.E., and Klumpp, J.H.: Analysis of Slip Damping With Reference to Turbine Blade Vibration, J. Appl. Mech., 23, 421 (1956). Shin, Y.S., Iverson, J.C., and Kim, K.S.: Experimental studies on damping characteristics of bolted joints for plates and shells, Trans ASME, J. Pressure Vessel Technology 113, 402-408 (1991). Nanda, B.K., Behera, A.K.: Study on Damping in Layered and Jointed Structures with Uniform Pressure Distribution at the Interfaces, J. Sound and Vibration, 226(4), 607-624 (1999). Ziada, H.H., Abd, A.K.: Load pressure distribution and contact areas in bolted joints, Inst. of Engineers (India), 61, 93-100 (1980). Damisa, O., Olunloyo, V.O.S., Osheku, C.A., and Oyediran A.A.: Static Analysis of Slip Damping With Clamped Laminated Beams. To Appear

APPENDIX
Notations
d dx x E I W b h

differential operator space coordinate modulus of rigidity moment of inertia static response width of Laminated beam depth of Laminated beam coefficient of friction at the interface of the laminated beams clamping pressure at the interface of the laminated beams applied end force amplitude length of Laminated beam shear stress at the interface of the laminated beams bending stress at the upper half of the laminates bending stress at the lower half of the laminates displacement of the upper laminate displacement of the lower laminate dummy variable static response in Fourier plane

P F L

xz (x)2 (x)1
u2 u1

WF

508
Objective Normalised Interface Pressure P = P0 Uniform Pressure Pav = P0 P = P0 (1+ x )
Pav = P0 1 + 2 P( x ) = P0 (sinh 1 x + B cosh 2 x )
Pav = P0 Pav =

V.O.S. Olunloyo, O.Damisa, C.A. Osheku and A.A. Oyediran


No slip (u = 0 ) Max energy Dissipation (D = Dmax )

P0 = 0.75 1 P0 = 0.75 1 H 1

P0 = 0.375 1 P0 = 0.375 1H 1

Pav = 0.75 1H 2

Pav = 0.375 1 H 2

(sinh x + B cosh x )dx


1 2 0

P0

1 2

[ 2 (cosh 1 1) + B1 sinh 2 ]
M

P( x ) = P0 am x m
m=0

Pav = 0.75 1H 3
M

Pav = 0.375 1 H 3

Pav = P0 am x mdx = P0
0 m =0

1 M

m =0

am m +1

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.509-522 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor
P. Mahanti University of New Brunswick, New Brunswick, Canada E-mail: pmahanti@unbsj.ca Mohammed Al-Fayoumi University of Graduate Studies,Amman, Jordan E-mail: mfayoumi99@yahoo.com Soumya Banerjee Birla Institute of Technology, Mesra, India E-mail: soumyabanerjee@bitmesra.ac.in Abstract We present ant colony based research Honeypots, a novel architecture that combines the merit of honeypot and intelligent anomaly prediction and prevention mechanism. It is observed that at higher levels variety of anomaly detectors are used to monitor all traffic to protect network service. Traffic susceptible to threat is processed by this proposed intelligent agent based technique, which is also capable to determine the accuracy of anomaly detection. Attacks or threats against these are trapped and their implications are nullified if happen. We contemplate a simulation of cyber attack towards honeypot, which attackers normally hate or avoid. Our goal is to present a model of honeypot where attackers are simulated through a colony of ant agents. The chemical mark up of ant agent's i.e. pheromone here identifies the presence of honeypot and thus probability of giving up to a particular attack happens. The end result of our simulation is significant enough to predict the future attacks (as the proposed honeypot by virtue can learn the pattern of attack) and parse this information to update the anomaly detection. In this proposed model honeypots utilize psychological and cognitive weapons such as deception or deterrence to confuse or stop attacks. Hence the psychology and motivation of attackers has been studied here. The paper also proposes a conceptual level of attack graph called as Honeypot attack graph (HAG) simulated through ant agents. Keywords: Honeypots, Ant Colony Metaphor, Attack Graph, Honeypot Attack Graph, Simulation

Introduction
In the context of demanding network security, the knowledge about the attack patterns and behavior of the attackers are extremely important parameters to prevent the possible threats across any enterprise network. The paradigm of network resources (terminals, switches, hubs and routers) are seriously probed and commissioned; they are contemplated as honeypots. Hence by definition a honeypot is a

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

510

security resource whose value lies in being probed, attacked, or compromised. The key point with this definition is honeypots are not limited to solving only one problem; they have a number of different applications. The term honeypot is usually refers to an entity with certain features that can lure hackers into its vicinity. These systems run special monitor programs which permanently help in data acquisition and substantially aids in post incident network scenario and forensics. The incorporation of electronic bait like honeypot is to capture more and more information about the activity of attacker, which in turn assist in detecting the possibilities of attack or helps in anticipating the cognitive probabilistic behavior of the hacker (Spitzner,2003) (Vrable et al, 2005). Hackers generally want to avoid honeypots for several interesting reasons. Firstly their activities are monitored and escaping would like to be difficult and they also don't like monitoring because they thrive on stealth and concealment for their efficiency. To better understand the value of honeypots, we can break them down into two different categories: production and research. Production honeypots are used to protect the network; they directly help secure the organization. Research honeypots are different; they are used to collect information. That information can then be used for a variety of purposes, such as early warning and prediction, intelligence gathering, etc. Considering these interesting features of honeypots, we envisage research honeypots not only to collect information or pattern of attack but subsequently we develop an algorithm for a specific type of deception mechanism comprising of ant colony optimization and pheromone probability matrix method. In this model of simulation attackers is ant colony. The motivation of using this technique is bi-fold, as already ant's pheromone simulation model successfully deployed several innovative applications. The method is attractive due to the collaborative sensing of ant agents. The architecture of the solution incorporated in functioning the honeypot physically and conceptually could detect the behavior of attackers. The rest of the paper is organized as follows: Section 2 enlists other significant works related to honeypots and creation of such honeypots. Section 3 describes the background introduction of the solution components i.e. ant colony optimization, pheromone probability in brief. Section 4 and its subsections discuss the specification and description of proposed algorithm of simulation for honeypot followed by a simple example in section 5. Section 6 provides conclusion, importance of such innovative solution and scope of future work.

Related Work- Honeypots and Creation of Honeypots


Lance Spitzner defines honeypots as follows (Spitzner, 2003): "A honeypot is an information system resource whose value lies in unauthorized or illicit use of that resource." A honeypot is a concept; a system built and set up in order to be hacked. It is rather the purpose that defines the system as a honeypot than any special program, configuration, etc. Its purpose can be either intrusion detection or attack research. A honeypot can be anything from a simulated service to a fully functional system. Since it is vulnerable, it should not be a production system; but in order to attract attackers it has to look like one. A honeypot must be visible and accessible for the attacker since, it is only useful if it is attacked. Otherwise no information can be deduced. On the other hand, no attacks on the honeypot do induce a secure network. To extract the crucial information various logging and sniffing techniques are available. There are not any rules as how a honeypot must be designed, but various guidelines for homemade honeypots exist. Whenever traffic to the honeypot is recognized, one can easily deduce an attacker's actions. Among the different functional honeypots the research honeypot (used in this work) has some special significance. When set up as a research honeypot it is a valuable system to analyze the attacker's actions. Because of the close monitoring, one can obtain a detailed view of the attacker's method of proceeding, his tools, the exploited vulnerability and the first steps after he has gained control of the honeypot. Detailed information about the attack is the foremost aim of the system, so logging and data capturing are much more detailed than in a production system.

511

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee

Honeypots can run any operating system and any number of services (Provos, 2005). The configured services determine the vectors available to an adversary for compromising or probing the system. A high-interaction honeypot simulates all aspects of an operating system. A low-interaction honeypots simulates only some parts, for example the network stack. A high-interaction honeypot can be compromised completely, allowing an adversary to gain full access to the system and use it to launch further network attacks. In contrast, low-interaction honeypots simulate only services that cannot be exploited to get complete access to the honeypot. Low-interaction honeypots are more limited, but they are useful to gather information at a higher level, e.g., learn about network probes or worm activity. They can also be used to analyze spammers or for active counter measures against worms when gathering information about network attacks or probes, the number of deployed honeypots influences the amount and accuracy of the collected data. A good example is measuring the activity of HTTP based worms (Song et al, 2001). We can identify these worms only after they complete a TCP handshake and send their payload. However, most of their connection requests will go unanswered because they contact randomly chosen IP addresses. A honeypot can capture the worm payload by configuring it to function as a web server. The more honeypots we deploy the more likely one of them is contacted by a worm. Each computer worm is different in the sense that each of them uses different mechanisms to spread on the Internet. Some of them just spread on a particular targeted network as opposed to the Internet. There are several features or parameters that characterize a worm such as: Vulnerabilities exploited on victim Speed of spread Strategies of spread Payload of the worm Intended goals and unintended effects (Senthilkumar,2004) Programmatically, in C language the worm has been declared as (Senthilkumar,2004): worm() { worm engine(); probe(); transport(); } worm engine() { install and initialize the worm(); execute payload if any needs to be(); choose the next host to probe(); wait until desired time for the next probe(); } Similarly, there is substantial impressive recent works related development of the Honeyd@WEB. Honeyd@WEB is a system that can deploy low-interaction, production, dynamic and manageable virtual honeypots via a web interface (Anuar et al, 2006). Even the popularity of medium interaction honeypots are also observed over different existing approaches to catch malware -especially bots. Medium Interaction Honeypots try to combine the benefits of both approaches in regards to botnet detection and malware collection while removing their shortcomings (Wichersk, 2006). While individual honeypots were becoming more reliable, the combinations of several such baits were tried. This was called as honeynet. A honeynet (Technical Report Section, Honeynet.org, 2006) (Spitzner, 2003) (Know your Enemy, Tech Report Series 2006) is a network of computers running honeypots. The inclusion of more smart security system slowly transformed the attackers more innovative in their approach. Therefore the aspect of system vulnerabilities was also in the limelight. One way to

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

512

visualize it was as deception. Fred Cohen's Deception Tool-Kit (DTK) (Fred Cohen t al, 2003) is a programmable toolkit of scripts designed to make it appear that a system contains a large number of well-known system vulnerabilities. DTK exposes the results of these vulnerabilities by the inputs of the unauthorized users' scans or attacks. The efficacy of Honeypots would be best if they can deceive attackers into thinking that they are normal computer systems. Most attackers do not want their methods known because they know that will lead to quick development of defensive methods to thwart them. So many attackers (including automated ones) will leave a computer system if they suspect it is a honeypot. For instance, many automated attacks look for the VMWare virtual-machine package on a system they are attacking, as it is often associated with recording their activities, and kill their processes if so. VMware allows multiple virtual machines, each running its own operating system, to co-exist on a single real machine. Potentially dangerous applications can thus be isolated from each other by running them in separate virtual machines. Since, the sophistication of attacks and deception mechanisms has taken different shapes, to combat all these designing and usage of honeypots also gets altered. This is reported in the recent work (Neil, 2006). The functionalities and the detection of suspicious environment are crucially important for preventing of any cyber attack. The pattern of attack, their comparison in real network with their known specification helps to develop a better efficient honeypot (Holz, 2005) (Duong, 2006). Another interesting and useful contribution has also been accomplished in most recent research to design a fake honeypot in order to deceive the attacker (Row, et al, 2006). Thus, the several initiatives adopted in recent cyber security paradigms clearly demonstrate the importance of honeypot and associated measures. In order to design more smart honeypots (or honeynet) it is obvious to have clear understandings related the creation and positioning of the honeypots. The actual honeypot consists of the followings : Transparent bridge Honeypot server Management and analysis system Each of these machines plays a crucial role in the security and reliability of the honeypot. Transparent bridge: It controls all traffic to and from the honeypot. In fact, it is the only real protection that we have between the honeypot and the network, and the Internet in general. Transparent bridging can also be accomplished with Linux and bridging software, but OpenBSD was more convenient at the time. Honeypot server: When building the honeypot it is very important that it can serve a number of different operating systems. It is also essential that a honeypot can be restored once it has been compromised. The obvious choice then is to use a VMware as a honeypot server. VMware can support a number of 'interesting' operating systems as guest, among which are Linux, Windows (NT, 2000 and XP) and FreeBSD. It runs on relatively simple hardware as long as there is enough memory and disk storage available (Donkers, 2002). Another related issue in designing and creating honeypot is to incorporate snorts (Caswell et al, 2001). The honeypot server could easily be equipped with snort, listening on the external interface, to perform real time alerting of break in attempts. It does mean that one has to maintain the rule set of snort on a very regular basis, and be able to act on these real time alerts. Furthermore, it can be argued that the honeypot can be used to detect new exploits that are not yet recognized by snort or any other IDS in the first place. It has been also suggested in research that Windows honeypots have a bigger problem compared to Linux or Unix platforms, since most key loggers work on the console and not on netcat based network enabled cmd sessions. They may use the NT root kit which provides some key logging functionality. Therefore in window driven scenario the user needs to install some program on the honeypot, so in this context the designer also runs the risk of being detected. It is therefore important to hide both this program and the data it generates.

513

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee

Foundation of Ant Colony Optimization


Ant Colony Optimization (ACO) (Dorigo, et al, 1999) is a recently proposed meta-heuristic approach for solving hard combinatorial optimization problems. The inspiring source of ACO is the pheromone trail laying and following behavior of real ants which use pheromones as a communication medium. In analogy to the biological example, ACO is based on the indirect communication of a colony of simple agents, called (artificial) ants, mediated by (artificial) pheromone trails. The pheromone trails in ACO serve as distributed, numerical information which the ants use to probabilistically construct solutions to the problem being solved and which the ants adapt during the algorithm's execution to reflect their search experience. Current applications of ACO algorithms fall into the two important problem classes of static and dynamic combinatorial optimization problems. Static problems are those whose topology and cost do not change while the problems are being solved. This is the case, for example, for the classic TSP, in which city locations and intercity distances do not change during the algorithm's runtime. Differently, in dynamic problems the topology and costs can change while solutions are built. An example of such a problem is routing in telecommunications networks (Caro et al, 1998), in which traffic patterns change all the time. The ACO algorithms for solving these two classes of problems are very similar from a high-level perspective, but they differ significantly in implementation details. The ACO metaheuristic captures these differences and is general enough to comprise the ideas common to both application types. The (artificial) ants in ACO implement a randomized construction heuristic which makes probabilistic decisions as a function of artificial pheromone trails and possibly available heuristic information based on the input data of the problem available for many combinatorial optimization problems.

Motivation Of Using Ant Colony Based Attack Graph and Pheromone Update
Applications of intelligent computing or heuristic computing in security prediction or alert have become prime topic of research and implementation since a decade. There are couples of conceptual and simulated applications proposed and developed for predicting intrusion detection in sensor network in recent times using either soft computing or ant colony optimization and swarm intelligence (Banerjee, et al, 2005) (Abharam et al, 2004, 2005). Subsequently, the substantial development of local vulnerability analysis encouraged the community to rely more on attack graph. Attack graphs can answer questions like "What attacks is my system vulnerable to?" and "From an initial configuration, how many different ways can an attacker reach a final state to achieve his goal?" or "What is the probability of give up by attackers"? Hence the monitoring of research honeypot, (which gathers information in different state) can be thought of an attack simulation graph based on ant colony agents. The choice of ant colony optimization is due to its distributed and collective intelligence through a chemical mark up called as pheromone. Pheromone update rules are available globally to reinforce or to check a particular path against a threshold and represent decisions. Thus the proposed innovative solution carries flavor of ant agent and attack graph model.

Proposed Bio-Inspired (Ant Colony) Algorithm, Positioning and Evaluation


The challenge in these systems from an intelligence viewpoint is to find a way to influence the opponent to concentrate their intelligence efforts against the honey pot over other systems. This challenge is relatively easy to meet against some classes of threats, such as typical Internet-based Web site defacers who look in advertised locations for high profile systems to attack. It is also effective as a system to 'switch' an attacker to once an attack attempt has been detected. Unfortunately, for even a mildly advanced intelligence activity, this type of defense offers little in the way of effective influence because it consumes such a small portion of the overall intelligence space and has little effect on altering the characteristics of the typical intelligence probe. Therefore, First we investigate the

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

514

mathematical model behind the attacker's behavior or reaction against our research honeypot driven by ant agent algorithm. We consider a manual attack where the hacker is progressing step by step separately. Obliviously, they are intelligent and can understand that they are targeting honeypots, where their patterns of actions are observed. Our model would like to demonstrate the probability value of attackers left over option if he is convinced that he is dealing with honeypot. In order to accomplish such feature in the proposed honeypots, we are keeping our ant agents in the honeypots or in honeynet architecture to simulate an attack towards most likely path and most recent threat by the updating their pheromone deposition rule.(Refer Fig. 1) The ant agents consult with an network attack graph to accomplish this simulation. The following parameters of standard attack graph in tune of ACO are considered: The next attack number or id The point of atomic attack The target id Pheromone value towards honeynet architecture The flagging of next attack will be detectable or stealthy. (Refer Proposed Algorithm) The ants in this case could discover the best feature combinations as they proceed throughout the search space, which is nothing but the traversal path under several honeypots. In nature, it can be observed that real ants are capable of finding the shortest route between a food source and their nest without the use of visual information and hence possess no global world model, adapting to changes in the environment. The deposition of pheromone is the main factor in enabling real ants to find the shortest routes over a period of time. Each ant probabilistically prefers to follow a direction rich in this chemical. The pheromone decays over time, resulting in much less pheromone on less popular paths. Given that over time the shortest route will have the higher rate of ant traversal, this path will be reinforced and the others diminished until all ants follow the same, shortest path. In this proposal, we are contemplating the ant agent simulating the attack or attackers behavior (typically probability value of attackers left over option) using ants global pheromone update rule.

Basic Mathematical Model and Derivation


In tune of initial assumption we make use of bio inspired network attack graph for predicting the probability of avoiding attack through our false honeypot. We find that to accomplish a complete functional model for designing a look alike or false honeypot using ant colony agents demand certain other assumptions as well (Popovsky, et al, 2006); although, these assumptions are supporting our proposed ant colony based network graph without intervening more with the core algorithm. Hacker's/attacker's motivation: M P is the probability of not failing to intrude P~ probability to give up (used in this paper) v value of success to the attacker c1 cost to the hackers c2 consequence We can write M = f [ (P (v) + P~ (v~)) (c1 +c2)] According to this support model hackers motivation is the summation of function of the probability equally not failing or to give up(inverse of success) to the attacker subtracted from the sum of costs and consequences. This support model motivates us to define a network attack model As the actual honeypot consists of the following machines: Transparent bridge

515

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee

Honeypot server Management and analysis system Therefore We identify the core S is a set of the combination of transparent bridge and Honeypot server connected to the network C Connecting topology in between T Trust relation expressing trust between hosts. I Model of intruder A attack scenario IDH model of the intrusion detection system at honeypots DH strategy of defense using honeypot We consider a scenario where there m potential attacks occurred by the attacker within G, which is basically is attack graph G = (S, E, s0, L) (Sheyner, et al, 2002, 2004) , Again, S is the set of attack states E S s is the set of edges,

s0 S is the initial state of attack and L : E { } is a labeling function


where L(e) = a if an edge e = (ss) corresponds an attack a otherwise L(e) = . It has been observed from the log data of honeypots that attackers repeatedly failing because they are committed to one approach (The Honey Net Project, 2004). Hence, the probability of giving up to a particular attack site depends on cost of reconnaissance factor as well. Therefore, we get (Sheyner, 2004) some mathematical expression as give up probability of attacker. Similarly, if xi defines the fraction of attacker(in this case its colony of ant agents) engaged in hacking a task i, em is the urge or factor of encouragement or motivation associated with task i without understanding that its honeypot, then we incorporate similar encouragement factor through the ant colony (which is reverse of give up) as pi is the probability of encouragement stimuli i, si is the importance of attack intensity and p is the give up probability and is the threshold value of chemical pheromone beyond which the attacker believes its honeypot then we propose a better attacking task and give probability and is the threshold value of chemical pheromone beyond which the attacker believes its honeypot then we propose a better attacking task and give up dynamics as :
s txi = pi ii 2
2

+ si

(1

xk ) pxi
K=1

This entire experimentation of network attack graph simulated through ant colony agents as attacker to believe the electronic bait as honeypot, finally led towards Honeypot Attack Graph comprising of tuple: HAG = (v, e, ph, v0, vs, , ) where, v is the set of vertices in a honeypot network graph E is the set of edges v0 is the set of compromised vertices of the honeypot network vs, is the set of target honeypot vertices from where the attacker will give up or leave the attack intention. : v ph is a function that drops pheromone to a vertices through ant agents :E ph is a function that assigns pheromone deposition cost to the edges and is the set of natural numbers. For example (v1) = ph that the amount of pheromone can be obtained at vertex v1, (e1) = (ph> ), where the threshold value of pheromone. [We are not including the cost incurred by the attacker if he deceives by the honeypot)

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

516

High Level Description of the Ant Based Honeypot Attack Graph Procedure AntAttackGraph (Attid, Nxtatt,TargetidHost, PhVal, NxtAttackType) /* Initialization */ Begin int Attid, Nxtatt, TargetidHost; Bool PhVal = FALSE; NxtAttackType = Dor S /* D: Detectable, S: Stealthy*/ For every S in the honeynet do /* S: S is a set of the combination of transparent bridge and Honeypot server Connected to the network*/ Normalize C to a unit vector in real N attack space /* C: Connecting topology in Between*/ Assign Ci to vi and place vi randomly on the honey net grid end for For every attack mi do Set ant agent at randomly selected Hotspot along the honeypot so that no two are Close; end for for all ants do Place ant at a randomly selected node and edge of attack graph G /* G = (S, E, s0, L)*/ If ((agent unladen) & (edge occupied by vector vi)) then compute f(vi) and pp (vi) ; draw a random real number R between 0 & 1 if (m pp(vi) then form path towards vi /*Probability of Attack path Formation in honeynet considering vertex vi*/ end if. else if (agent carrying item vi) & (honeypot site) then compute f(vi) and pp (vi) ; draw a R. if (m pd(vi)) then drop item vi; endif end if If (agent carrying them vi) then for every ant Ai do Evaluate distance (mi, X) /* Euclidian distance in N space attack graph*/ end for. Select Attack m for which the distance is minimum & Attid, Nxtatt, TargetidHost = VALUE; For every Ai & mi in N space neighborhoods of G do Set m = [f P (v) + P~ (v~)]; if NxtAttackType == D or S then Update the size of pheromone matrix / *Pheromone Trail updating*/ End for. End if Update the size of pheromone matrix / *Pheromone Trail updating*/ e [ant] = 0 /* energy of ant*/ for every coupling (i,j) do update pheromone trails by applying the rule

517

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee ij (t) (1- ). ij (t) + ij (t) where 1/ c( + ) ij (t) = otherwise 0 end for Set PhVal = TRUE;

According to ant movement rule: /* ( r , u ) is the intensity of pheromone on the edge between the node position r and u, and

( r , u ) * ( r , u ) p= k (r , u ) * (r , u )

represents a weight for ( r , u ) */ Refer [Dorigo, 2000] */

the pheromone and weight for the heuristic of

for every coupling (i,j) do ij (t +1) = ij (t) Evaporate pheromone by k at all grid & evaluate give up probability value of attack according to
s txi = pi ii 2
2

+ si

(1

xk ) pxi

/* Refer equation (1)*/

For all ants do Decrease the energy e [ant] = e [ant] - e If e [ant] < 0.0 then Kill the ant and destroy attack path towards vi End if Print pheromone distribution and consult G end for. /* value of , and are finally used in section 5*/

Technical Specification for Implementation


In this work we are concentrating on targeted attacks. Targeted attacks hinge on the classification of one point or a small set of points. They are more sensitive to variations in the decision boundary than indiscriminate attacks because boundary movement is more likely to change the classification of the relevant points. The analyzation process could be engineered. A honeypot has to hide in the environment to look inconspicuous. It has to be checked, whether a honeypot would look more suspicious if it is the only host that does not have an IDS protection. The logging component is integrated into the Firewall host. Ideally the firewall is used for egress traffic filtering, and provides a gateway for the honeynet environment. It is designed to block any unestablished outbound connections. Inbound traffic is allowed to any of the public IP addresses we have reserved for the targets. This traffic is then translated to our local network (via NAT) and then forwarded to the appropriate target. Return traffic is handled in the same fashion. At the same time, copies of all traffic that pass through the switch is forwarded through the monitor port through a read-only cable to our intrusion detection system. This lets our HAG monitor all the traffic passing through the network, while still being relatively well-protected from outside interest. The targets themselves are VMWare guests running on a Windows host. We are using VMWare hosts for the targets because it is easy to maintain, back-up, and store the state of a VMWare guest. Each physical host runs two guests, and we have two hosts, giving us four targets in total. On

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

518

each of the VMWare guest targets, we are running honeyd, which is a low interaction honeypot. Significantly, this low interaction honeypot is converted to research honeypot when we conceptualize Honeypot Attack Graph. The code snippet will look like: typedef struct Metadata { struct sockaddr_in client; // the attacker struct sockaddr_in server; // the honeypot pid_t uid; // process data of mitm child pid_t gid; pid_t pid; pid_t ppid; } Mitm is the main class of mitm. Mitm initiates a Connector as a control channel and connects to the monitor. Then it creates a Reactor and adds the Event Handlers (WinAPI calls)
Figure 1: Honeypot Simulation through Ant Agents

Example Analysis
The simulation of ant based attack graph is achieved considering Window based honeypot. The time versus pheromone deposition towards honeynet has been evaluated as shown in Fig. 2 and Fig. 3. Both these graphs demonstrate that under different time slice typical window based honeypot architecture becomes successful in keeping attackers away from the attack point. The recoding of pheromone value and attributes of attack graph also consolidate that the simulation is useful to model successful designing honeypot or honeynet architecture irrespective of Linux and Window compatibility.

519

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee


Figure 2: Honeypot Detection Using Pheromone

Simulation of Ant Colony based Attack graph


10

Honeypot Attck Graph

P h e r o m o n e D e p o s it io n

0 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 10

Time

Honeypot

Attackers Give up point against

Figure 3: Bypassing Attacker

Bypassing Honeypot of Attackers

Conclusion and Further Scope


Since last couples of years honeypot have been used in wide special applications like to provide traps for email or worms [Riordan, et al, 2005], and can be generalized to "honeytokens" Or individual

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor

520

pieces of disinformation [Spitzner, 2005]. Still traditionally it deceives the attacker and protects the network from suspected attack. Keeping these aspects in front the intelligent and heuristics component have been included and a simulation has been performed through ant colony. The coordination of ant agent and their remote relationships have been observed against basic honeypot design. An intense analysis of the State-of-the-Art led to the conclusion that honeypots could be improved to catch-up new attacks. Current systems depict that any traffic directed to the honeypot can be considered an attack. The tests showed that honeypots are exposed to more than just attack traffic. Current security labs are confronted with the problem that they get too much unfiltered reports about attacks. The usage of automated attack analyzing honeypots like this ant colony based Honeypot Attack Graph can be a step into the right direction. In this paper we have developed an algorithm of ant based attack graph to exhibit the functionality of honeypot and in coming days we would like to simulate more such behavioral patterns of hackers or attackers using distributed intelligence and agent coordination so that sound alert and security analysis can be accomplished.

521

P. Mahanti, Mohammed Al-Fayoumi and Soumya Banerjee Anuar , N. B. , Zakaria , O., and Yao , C.W. : Honeypot through Web (Honeyd@WEB): The Emerging of Security Application Integration, Issues in Informing Science and Information Technology, Volume 3, (2006) Abraham, A., Jain, R., Thomas. J., Han, S. Y. , : D-SCIDS: Distributed soft computing intrusion Detection system. Journal of Network and Computer Applications, Elsevier Ltd. (2005) pp. 1 -19. Abraham, A., Jain, R., Sanyal, S. , Han, S. Y. :SCIDS: a soft computing intrusion detection system. Sixth international workshop on distributed computing (IWDC 2004) Lecture Notes in Computer Science, vol. 3326 Germany: Springer; 2004. p. 25270 ISBN: 3-540-24076-4 Abraham, A., Thomas, J. , : Distributed intrusion detection systems: a computational intelligence approach. In: Abbass HA, Essam D, editors. Applications of information systems to homeland. Security and defense USA: Idea Group Inc. Publishers; 2005. pp. 10535 [Chapter 5] Banerjee, S., Grosan, C., Abraham, A., Mahanti, P. K. (2005) : Intrusion detection on sensor networks using emotional ants. Int. J. of Applied Science and Computations, USA. Caswell,B. and Hewlett, Jeremy. Snort Users Manual Snort: http://www.snort.org Caro, G. D. and Dorigo, M., : AntNet: Distributed stigmergetic control for communications networks. Journal of Artificial Intelligence Research, 9:317365, (1998) Dorigo, M. and Caro, D. G. The Ant Colony Optimization meta-heuristic. In D. Corne, M. Dorigo, and F. Glover, editors, New Ideas in Optimization, McGraw Hill, London, UK, (1999), pp.1132 Dorigo, M., Caro, G. D., and Gambardella, L. M., : Ant algorithms for discrete ooptimization. Artificial Life, 5(2): (1999) pp.137172. Donkers, Arthur, Honey, I caught a worm, 3rd International SANE Conference, MECC, Maastricht, The Netherlands, May 27 - 31, 2002. Duong, B., : Comparisons of Attacks on Honeypots with Those on Real Networks, M.S. Thesis, Naval Postgraduate School, www.cs.nps.mil/people/faculty/rowe/oldstudents/duong_thesis_final.htm, March (2006). Dorigo, M. : Universite Libre de Bruxelles, IRIDIA,The Ant Colony Optimization Metaheuristic: Algorithms, Applications, and Advances, Technical Report IRIDIA-2000-32. http://Honeynet.org, Technical Report Section Holz , T. and Raynal , F. , : Detecting Honeypots and other suspicious environments, Proceedings of the 2005 IEEE Workshop on Information Assurance and Security, United State Military Academy, West Point, NY, 15-17 June (2005), pp.1-8, ISBN 0-7803-8572-1. Know Your Enemy: Honeynets, www.honeynet.org/papers/honeynet/ index.html Neil, C. R., Measuring the Effectiveness of Honeypot Counter-Counterdeception, proceedings of the 39th Hawaii International Conference on System Sciences (2006), pp. 1-10 ISBN 07695-2507-5/06. Provos, N.,: A Virtual Honeypot Framework, Google, Inc.(2005). Popovsky, B. E. and Frincke, D. , : Adding the Fourth "R": A Systems Approach to Solving the Hacker's Arms Race, Proceedings of the Symposium 39th. Hawaii International Conference on System Sciences, 2006 pp. 1-10. Rowe, N.C., Duong, B.T. and Custy, E.J. , : Fake Honeypots: A defensive Tactic for Cyberspace, appeared in Proceedings of the (2006) IEEE Workshop on Information Assurance and Security, United State Military Academy, West Point, NY, 21-23 June (2006) Riordan, J., Zamboni, D., and Duponchel, Y., "Lessons Learned from Billy Goat, an Accurate Worm-Detection System," Technical Report RZ3609, IBM Research GbmH, Zurich Research Laboratory, May 16, (2005). Song, D., Malan , R. , and Stone , R . : A snapshot of Global Worm Activity. Technical Report, Arbor Networks, November (2001)

References
[1] [2] [3] [4]

[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

Simulating Targeted Attacks Using Research Honeypots Based on Ant Colony Metaphor [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31]

522

Spitzner, L.,: Tracking hackers. http://www.tracking-hackers.com/, (2003) Spitzner, L., "Honeytokens: The Other Honeypot," Retrieved May 30, (2005), from www.securityfocus.com/infocus/1713. Senthilkumar, G. C.,: Modeling a Computer Worm Defense System, MS Thesis Report, University of California, Davis, (2004), pp.1-89. Sheyner, O., Haines, J., Jha, S., Lippman, R., and Wing, J. M., on Automated Generation and Analysis of Attack Graphs, Proceedings of the IEEE Symposium Security and Privacy, Oakland, CA, May 2002. Sheyner, O. , Scenario 04- Graphs and Attack Graphs, CMU Computer Science Department technical report CMU- CS-122, Ph.D. dissertation, April 2004. Sugerman, J., Venkitachalam, G. and Lim, B. H. : Virtualizing I/O Devices on VMware Workstations Hosted Virtual Machine Monitor. In Proceedings of the Annual USENIX Technical Conference, June (2001) pp.530. The Honeynet Project, Know Your Enemy: Learning about Security Threats, Second Edition, Boston, MA: Addison-Wesley, (2004); Wicherski , G., : Medium Interaction Honeypots April 7, (2006); www.all.net: Fred Cohen & Associates' ForensiX. Vrable, M. , Ma, J. ,Chen, J., Moore, D., Vadekieft, E., Snoeren, A., Voelker, G., and Savage, S., : Scalability, Fidelity, and Containment in the Potemkin Virtual Honeyfarm, Proc. Symposium on Operating Systems Principles, Brighton UK, 2005, pp. 148-162

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.523-531 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Intrusions Detection Mechanism by Resilient Back Propagation (RPROP)


Iftikhar Ahmad Department of Computer Sciences COMSATS Institute of Information Technology, Abbottabad, Pakistan Sami Ullah Swati Department of Computer Sciences COMSATS Institute of Information Technology, Abbottabad, Pakistan Sajjad Mohsin Department of Computer Sciences COMSATS Institute of Information Technology, Abbottabad, Pakistan Abstract The accurate detection of computer and network system intrusions has always been an elusive aim for system administrator and information security researchers. There are several methods of responding to a network intrusion, but they all require precise identification of the attack. This paper presents an approach of intrusion detection by Resilient Backpropagation algorithm, a direct adaptive method for faster learning. Keywords: Intrusion Detection, Anomaly Detection, Misuse Detection, RPROP (Resilient Backpropagation), Neural Networks, IDS (Intrusion Detection System), Learning Algorithm, Dataset.

1.0. Introduction
Intrusion detection systems are the main tools for providing security in computer and network system. The dependency of private and government organizations is increasing on their computer networks and protecting theses system from attack is serious. A single intrusion of a computer network can result in loss or the reliability of network became unsure. Therefore a mechanism is presented here that will detect network intrusion in a precise way by using the power of RPROP learning algorithm [1]. This paper is divided into three primary areas. The first section provides an overview of intrusion detection and neural networks. The second section presents our architecture of neural networks in the identification of attacks against a network. Finally, the results of tests conducted on a neural network, which was designed as a proof of concept, are also presented and the areas of future research that are being conducted in this area are also discussed.

2.0. Intrusion Detection


Intrusion Detection Systems (IDSs) are the most important tools for providing security in computer and networks. There are numerous methods of responding to a network intrusion, but they all require the precise and timely identification of the attack [2]. Dr. Dorothy Denning proposed an intrusion

Intrusions Detection Mechanism by Resilient Back Propagation (RPROP)

524

detection model in 1987 which became a landmark in the research in this area. The model which she proposed forms the fundamental core of most intrusion detection methodologies in use today [3]. The intrusion diction systems can be classified into three categories as host based, network based and vulnerability assessment [4]. There are many approaches that are being used to accomplish the desirable elements of an intrusion detection system like anomaly detection, misuse detection, combined anomaly/misuse detection, pattern recognition and networking monitoring. 2.1. Neural Networks An artificial neural network consists of a collection of processing elements that are highly interconnected and transform a set of inputs to a set of desired outputs. The result of the transformation is determined by the characteristics of the elements and the weights associated with the interconnections among them. By modifying the connections between the nodes the network is able to adapt to the desired outputs [5, 6]. The application of neural network in intrusion detection is an on going area. It offers the potential to resolve a number of the problems encountered by the other current approaches to intrusion detection. Artificial neural networks are alternatives. The first advantage in the utilization of a neural network in the intrusion detection would be the flexibility that the network would provide. . A neural network would be capable of analyzing the data from the network, even if the data is incomplete or distorted. Similarly, the network would possess the ability to conduct an analysis with data in a non-linear fashion. Further, because some attacks may be conducted against the network in a coordinated attack by multiple attackers, the ability to process data from a number of sources in a nonlinear fashion is especially important. The natural speed of neural networks is another benefit of this approach. However, the most important advantage of neural networks in intrusion detection is the ability of the neural network to "learn" the characteristics of attacks and identify instances that have been observed before by the network.

3.0. Architecture
The architecture of our mechanism consists of one input, two hidden and one output layer with 41,14,9 and 2 neurons. The input layer consists of 41 neurons because the Kddcup data set contains 41 fields for a network packet to be used for intrusion detection training [7, 8, 9, 10]. There is no exact formula for the selection of hidden layers so we can make it by comparison and select which one is best [11]. For choosing best set of hidden layers and its no. of neuron a comparison is made for many cases and best one is selected that is two hidden layers with 14 and 9 neurons. The architecture for normal and attacked scenario is shown.
Figure 1

525

Iftikhar Ahmad, Sami Ullah Swati and Sajjad Mohsin

The architecture used in our mechanism is feedforward. If the output of neural network is nearest to 0 and 1 then this packet will be considered as normal.
Figure 2

If the output of neural network is nearest to 1 and 0 then this packet will be considered as bad means it involves some intrusion. Our mechanism consists of two phases. One is training of the net while second is testing of the net. The training phase is shown in the following figure.
Figure 3

The training architecture consists of layers. The Layer contains units or nodes called neurons. A neuron is a processing element that has an activation function to produce its output. The layers of the net are interconnected by synapse. The Layer object is the basic element that forms the neural network. It is composed of neurons, all having the same characteristics. This component transfers the input pattern to the output pattern by executing a transfer function. The output pattern is sent to a vector of Synapse objects attached to the layer's output. It is the active element of a neural net. The Layer applies a sigmoid transfer function to its input patterns.

Intrusions Detection Mechanism by Resilient Back Propagation (RPROP) Input layer

526

The Input layer takes input from the input file that contains dataset for training of the net. The sample dataset is randomly selected from Kddcup dataset. The 80% from the samples are used for training while others are used for testing the net. The row of the dataset is called a pattern representing an instance of the input dataset. The neural network reads and elaborates sequentially all the input rows and for each one it generates an output pattern representing the outcome of the entire process. Hidden layers We use two hidden layers in our developed mechanism representing a good non-linear element of the neural network. The sigmoid layers are used as hidden layers of the net. It can also be used to build whatever layer of a neural network. These take inputs from the outputs of the input layer and apply its activation function. The output is sent to the output layer. Output layer The Output components allow a neural network to write output patterns in a file that are used for analysis of intrusion. Teacher layer The function of this component (the TeacherSynapse) is to calculate the difference between the output of the neural network and a desired value obtained from some external data source. The calculated difference is injected backward into the neural network starting from the output layer of the net, so each component can process the error pattern to modify the internal connections by applying the learning algorithm. The Synapse The Synapse represents the connection between two layers, permitting a pattern to be passed from one layer to another. The Synapse is also the memory of a neural network. During the training process the weigh of each connection is modified according the learning algorithm. The layers are fully connected with each other. For this purpose FullSynapse object is used that connects all the nodes of a layer with all the nodes of the other layer, as showed above in the figures: Training We used Resilient Backpropagation algorithm for training of the net because it converges very quickly. It uses only the sign of the backpropagated gradient to change the biases/weights of the net. Hence it is local adaptive learning scheme and has the possibility to escape from local minima. The basic principle is to eliminate the harmful influence of the size of partial derivative on the weight step [12, 13, 14, 15]. The training phase consists of the following steps. 1. The Feedforward of input training pattern 2. The calculation and backpropagation of associated error 3. The adjustment of the weights The aim of training means to get reasonable (good) responses to input that is similar but not identical. Testing After training, the net only involves computation of the feedforward phase. The structure of the testing net is shown in the figure.

527

Iftikhar Ahmad, Sami Ullah Swati and Sajjad Mohsin


Figure 4

It consists of one Input, two hidden and one output layers. The output of the net is saved in file that will be used for intrusion analysis. If the global error value is nearest to 0 and 1 then it is normal packet otherwise consider as attack. The net is implemented in Java by using JOONE API. The interface of the simulation is shown in the figure.
Figure 5

The summary of report generated by the simulation is shown in the following figure. Summary: No. of Total Packets= 16 No. of Normal Packets= 5 No of Attacked Packets= 11

Intrusions Detection Mechanism by Resilient Back Propagation (RPROP)


Figure 6

528

The different reports are generated by the simulation for several attacks and then find out its accuracy.

4.0. Results
After training, the net is tested for different types of network attacks. A comparative analysis of various attacks is shown in the following tables.
Table 1: Attack Category: DOS
Original Nrl 5 5 2 5 9 4 Atk 11 8 10 10 10 3 Nrl 5 5 2 4 8 1 Tested Atk 11 8 10 11 11 6 Detection Rate 100% 100% 100% 99% 99% 50%

Attack Type Back Land Neptune Pod Smurf Teardown

Table 2:

Attack Category: Probing


Original Nrl 10 2 4 4 Atk 3 2 5 9 Nrl 10 2 4 4 Tested Atk 3 2 5 9 Detection Rate 100% 100% 100% 100%

Attack Type IPsweep Nmap Portsweep Satan

Table 3:

Attack Category: R2U


Original Nrl 7 3 4 2 4 Atk 7 3 3 2 6 Nrl 5 3 3 2 4 Tested Atk 9 3 4 2 6 Detection Rate 77% 100% 100% 100% 100%

Attack Type Guespswd Imap Multihop Phf warezmaster

529
Table 4: Attack Category: U2R
Original Nrl 4 2 2 Atk 4 3 4

Iftikhar Ahmad, Sami Ullah Swati and Sajjad Mohsin

Attack Type Buffer overflow Perl rootkit

Tested Nrl 4 2 2 Atk 4 3 4

Detection Rate 100% 100% 100%

Table 5:

Attack Category: extra


Original Nrl 5 0 0 6 Atk 4 10 15 12 Nrl 5 0 4 6 Tested Atk 4 10 11 12 Detection Rate 100% 100% 73% 100%

Attack Type Httptunnel Mailbomb Mscan Procestable

The comparative results shows that it detection rate for DOS attack is 91.0 %, R2U is 95.4 %, U2R and probing is 100 % while it also detected some extra attacks that is 93.25 %.The overall detection rate is 95.93 %.

5.0. Conclusion
Our implemented mechanism shows best performance as compared to other NN approaches as shown in the following figure [7, 16].
Figure 7

Com parison in performance


100 Detection 95 rate 90 85 93.5 89 MSOMS ART-1 90.7 95.93

ART-2 RPROP

NN IDS Approches

The focus of our research was to increase accurate intrusion detection rate. Once attack is detected then there are several available methods to block network intrusion.

6.0. Future Research


The efficient intrusion detection mechanism can be developed that have very low error rate, high learning rate and quick intrusion detection by using with other neural networks in the form of hybrid architecture.

Intrusions Detection Mechanism by Resilient Back Propagation (RPROP)

530

References
[1] [2] [3] [4] [5] [6] [7] M. Riedmiller and H. Braun, A direct adaptive method for faster backpropagation learning: The RPROP algorithm, Proc of the IEEE Int. ConJ On Neural Networks, San Fransisco, 1993. Cannady J. Artificial neural networks for misuse detection. National Information Systems Security Conference; 1998. p. 36881. Denning, Dorothy. (February, 1987). An Intrusion-Detection Model. IEEE Transactions on Software Engineering, Vol. SE-13, No. 2. Jean-Philippe,Information Security, SANS Institute,2001. Hammerstrom, Dan. (June, 1993). Neural Networks At Work. IEEE Spectrum. pp. 26-53. Fox, Kevin L., Henning, Rhonda R., and Reed, Jonathan H. (1990). A Neural Network Approach Towards Intrusion Detection. In Proceedings of the 13th National Computer Security Conference. M. Amini , R. Jalili, Network-Based Intrusion Detection Using Unsupervised Adaptive Resonance Theory (ART), Fourth International ICSC Symposium on ENGINEERING OF INTELLIGENT SYSTEMS (EIS 2004) in collaboration with the University of Madeira Island of Madeira, Portugal February 29 March 2, 2004. The 3rd International Knowledge Discovery and Data Mining Tools Competition, http://kdd.ics.uci.edu/ databases/kddcup99/kddcup99.html, 2003. M. Sabhnani and G. Serpen, "Application of Machine Learning Algorithms to KDD Intrusion Detection Dataset within Misuse Detection Context", http://www.eecs.utoledo.edu/~serpen/professional/Research/Publication/MLMTA 2003 Manuscript Submission Version.pdf, July 2003. Aleksandar Lazarevic, Levent Ertoz, Vipin Kumar, Aysel Ozgur, Jaideep Srivastava, A Comparative Study of Anomaly Detection Schemes in Network Intrusion Detection,http://www.siam.org/meetings/sdm03/, SIAM International Conference on Data Mining (2003). www.ieeexplore.ieee.org/iel3/1723/4574/00181220.pdf JJF Cerqueira, AGB Palhares, MK Madrid , Man and Cybernetics, 2002 IEEE International Conference on, 2002. ELECTRONICS LETTERS 8th January 2004 Vol. 40 No. 1. Souza, B.A.; Brito, N.S.D.; et al, Comparison between backpropagation and RPROP algorithms applied to fault classification in transmission lines, S.S.B. Neural Networks, 2004. Proceedings. 2004 IEEE International Joint Conference on Volume 4, Issue, 25-29 July 2004 Page(s): 2913 2918 vol.4. L. Jiao et al. (Eds.): ICNC 2006, Part II, LNCS 4222, pp. 364 373, Springer-Verlag Berlin Heidelberg (2006). Rhodes, B., Mahaffey, J., & Cannady, J. (2000, October). Multiple Self-Organizing Maps for Intrusion Detection. Proceedings of the 23rd National Information Systems Security Conferenc

[8] [9]

[10]

[11] [12] [13] [14]

[15] [16]

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.531-546 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton
Naqib Ullah Khan Visiting Scholar, Department of Horticulture and Crop Science 1680 Madison Ave., OARDC, Ohio State University, Wooster, OH 44691, U.S.A NWFP Agricultural University Peshawar 25130, Pakistan E-mail: nukmarwat@yahoo.com; khan.213@osu.edu Tel: 001-330-202-3595; Fax: No. 001-330-263-3887 Gul Hassan NWFP Agricultural University Peshawar 25130, Pakistan Moula Bux Kumbhar Sindh Agriculture University Tandojam, Pakistan Sungtaeg Kang Visiting Scholar, Department of Horticulture and Crop Science 1680 Madison Ave., OARDC, Ohio State University, Wooster, OH 44691, U.S.A National Institute of Crop Science, RDA 151 Seodung-Dong, Suwon, 441-857, Korea Tel: 001-330-202-3595; Fax: No. 001-330-263-3887 Ikramullah Khan NWFP Agricultural University Peshawar 25130, Pakistan Aisha Parveen Graduate Students, Department of Plant Breeding and Genetics NWFP Agricultural University Peshawar, Pakistan Umm-E-Aiman Graduate Students, Department of Plant Breeding and Genetics NWFP Agricultural University Peshawar, Pakistan Abstract Performance of F1 hybrids, F2 population and some selected population in advanced generation (F3, F4 and F5) were studied for bolls per plant, seed cotton yield per plant, lint % and staple length of Gossypium hirsutum during 1999-2004 at Agricultural Research Institute, Dera Ismail Khan, NWFP Pakistan. In F1 hybrids, maximum heterosis was observed for seed cotton yield per plant followed by bolls per plant while lint % and staple length showed low values of heterosis. Majority of the F2 population displayed inbreeding depression and it was high for seed cotton yield per plant followed by bolls per plant, lint % and staple length. Inbreeding depression was higher in high performing hybrids than in low performing hybrids. In F1 generation, heterosis over better parent ranged from +3.13 to +65.63% for bolls per plant, +0.82 +115.22% for seed cotton yield per plant, +0.27 +3.88%

532

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman for lint % and +0.28 to +6.00% for staple length. In F2 population the average some selections were made in promising cross families (CIM-109 x CIM-1100, CIM-240 x CIM1100, CIM-1100 x CIM-109 and CIM-1100 x CIM-240). In advanced generations (F3, F4 and F5), the single plant families of the above four selected crosses superseded the standard cultivar (CIM-446) for bolls and seed cotton yield per plant and staple length, but in lint % only few of the cross families exhibited superiority over standard. Keywords: Upland cotton; F1 hybrids; F2 population; Mean Performance; Heterosis and Inbreeding Depression; Segregating Generations.

Introduction
Heterosis works as a basic tool for improved production of crops in the form of F1 hybrids. The phenomenon of heterosis of F1 hybrids can reflect their own specific combining ability (SCA) and the general combining ability (GCA) of parental lines. Hence, heterotic studies can provide the basis for exploitation of valuable hybrid combinations and their commercial utilization in future breeding programmes. Cook (1909) was the first to utilize hybrid vigour in interspecific hybrids (G. barbadense x G. hirsutum) and after that a number of workers all over the world are still supporting his contention. Hybrid cotton is a good approach for significant improvement in genetic potential for yield and fibre quality traits and has attracted attention of cotton breeders for commercial growing of hybrid generations. However, the efforts have not delivered the expected results due to its self pollination which has some different implications on hybrid seed production in comparison to cross pollinated crops (Waller et al. 1981; Waller and Moffett, 1981; Salam, 1991 and Arshad, 1991). In countries, however, like India and China, where labour is cheaper, the successful hybrid cotton produced on large scale since 1960s as reported by Krishnaswami and Kothandaramon (1977), Salam (1991), Soomro et al. (1991) and Khan et al. (1999a). In view of the economic importance of hybrid cotton cultivation, and the importance of selected hybrids in segregating generations, a research project was undertaken to evaluate the manifestation of heterosis in F1 hybrids and inbreeding depression in F2 population for bolls per plant, seed cotton yield per plant, lint % and staple length and also to study the other segregating generations for selected breeding material in upland cotton in a set of 6 parent diallel cross.

Materials and Methods


The crossing block, F1 and F2 population and experiments comprising of other advanced generations (F3, F4 and F5) of upland cotton (Gossypium hirsutum L.) have been conducted and maintained at Agricultural Research Institute, Dera Ismail Khan, Pakistan during 1999-2004. Six diverse genotypes of upland cotton i.e. CIM-109, CIM-240, CIM-1100, FH-682, BH-36 and CRIS-9 were sown during May 1999, in a non-replicated crossing block by dibbling on a well prepared seed bed. Five rows, each of 27 meters length, at the spacing of 60 and 100 cm between plants and rows, respectively, to ensure easy crossing and to handle the breeding material carefully. All cultivars were crossed in a complete diallel fashion. The F1s were raised during 2000 from half of the crossed seed to get F2 seeds. The F1 and F2 population of 6 X 6 complete diallel cross, having thirty hybrids (including reciprocals) along with six parents were sown by dibbling with randomized complete block design on a well prepared seed bed during May 2001. In F1 each genotype was planted in a single row measuring 3.30 m, with three replications, while in F2 the plant population was increased and each genotype was planted in four rows, each of 6.30 m length, with four replications. The row and plant spacing were 75 and 30 cm, respectively. All cultural practices were carried out as per recommended package for cotton production and all the crops were maintained under uniform conditions to discourage environmental variability to

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton

533

the maximum possible extent. Picking was made in November-December every year on single plant basis and ginning was done with eight saw-gin. Single plant selections were made in F2 crosses on the basis of cotton leaf Curl virus (CLCV) resistance, yield and yield components, which were sown with a standard cultivar (CIM-446) for other segregating generations i.e. during May 2002 for F3, May 2003 for F4 and May 2004 for F5. Each single plant family consisted of four rows of 5 meter in length with spacing of 75 and 30 cm. All cultural practices were carried out as recommended for cotton production and all crops were maintained under uniform conditions to the maximum possible extent to discourage environmental variability. Picking was made in November-December every year on single plant basis and ginning was done with eight saw-gin. In all the experiments the data were recorded for bolls per plant, seed cotton yield per plant and after ginning for lint % and staple length. The staple length was studied by High Volume Instrument (HVI) USTER-900A. The data were subjected to ANOVA technique using MSTATC, a computer software package (Bricker, 1991) for all the traits to test the null hypothesis of no differences between various F1s as well as among F2s hybrid populations and their parental lines, while in advanced generations the means of single plant families were compared with standard cultivar (CIM-446) for their percent increase or decrease. In F1 hybrids, heterosis was estimated according to Fehr (1987). The F1 hybrid means having positive heterotic values were further tested for significance with t test of Wynne et al. (1970). The observed inbreeding depression in F2 hybrids was calculated as percent decrease of F2 hybrids when compared with F1 hybrids as outlined by Hallaner and Miranda (1986) and Baloch et al. (1993a).

Results
The null hypotheses for equality of 30 F1 and F2 means and their six parental lines tested through ANOVA (Table 1), proved false at P0.01 for all the parameters and manifested highly significant mean differences. The results are presented here in seriatim. Boll weight The boll weight varied from 2.38 g (CIM-109) to 2.91 g (CIM-1100) among the parents, while CRIS-9 showed maximum boll weight (2.91 g). Among the F1 hybrids boll weight ranges from 2.42 g (CIM109 x CRIS-9) to 3.62 g (CRIS-9 x CIM-1100) (Fig. 1). Nearly 2/3 of F1 hybrids surpassed the parental cultivars. The highest and statistically equal boll weight was recorded in 9 F1 hybrids and their reciprocals in which eight crosses consisting of parent CIM-1100 and their boll weight was ranged from 3.34 to 3.62 g. The lowest boll weight was manifested by two cultivars CIM-109 (2.38 g) and CRIS-9 (2.52 g) and their F1 hybrid and its reciprocal having 2.42 to 2.46 g boll weight. The cross CIM-240 x CRIS-9 (2.61 g) also possessed the least boll weight. In F2 population, boll weight varied from 2.26 g (CIM-109 x BH-36) to 3.20 g (CIM-1100 x FH-682) among the hybrids (Fig. 1). The highest and statistically equal boll weight ranged from 2.96 to 3.20 g was recorded in 6 genotypes and reciprocals, which have CIM-1100 as one parent as also observed in F1 generation, The lowest boll weight was manifested by the genotype CIM-109 x BH-36 (2.26 g). The heterobeltiotic values ranged from +0.75 (BH-36 x CIM-240) to +24.40% (CRIS-9 x CIM1100) over better parents, respectively (Fig. 1). In case of heterobeltiosis 13 F1 hybrids surpassed their better parent with highly significant values and only 2 crosses showed significant heterobeltiosis. All the F2 genotypes displayed inbreeding depression for boll weight (Fig. 1) and the observed inbreeding depression was -0.37 to -23.66%. However, some genotypes displayed positive values of inbreeding due to transgressive recombination in F2 generation. The results showed that in F2 even after inbreeding depression, some promising population exhibited good performance and positive selection in such crosses can provide better base for further improvement. Due to high inbreeding depression for

534

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

boll weight as compared to other yield related traits, the selection for yield in F2 generation is mostly dependent on boll weight and bolls per plant, which ultimately control the seed cotton yield.
Table 1: ANOVA and CV% for boll weight, bolls per plant, seed cotton yield, lint% and staple length in F1 and F2 generations in upland cotton.
Parameters Boll weight Bolls per plant Seed Cotton Yield Lint percentage Staple Length
** significant at P0.01.

F1 F2 F1 F2 F1 F2 F1 F2 F1 F2

Reps. 0.059 0.034 10.333 3.130 13.194 39.719 0.573 1.149 0.851 1.355

Mean Squares Genotypes 0.404 0.170 261.819 83.140 4472.994 1343.963 4.773 4.560 5.262 3.982

F. Ratio Error 0.037 0.033 3.819 5.777 25.337 28.560 0.745 0.282 0.454 0.584 10.90** 5.23** 68.56** 14.39** 176.54** 47.06** 6.41** 16.19** 11.59** 6.82**

CV % 6.42 6.70 5.52 7.50 4.70 6.20 2.47 1.55 2.47 2.79

Figure 1: Mean performance, heterosis and inbreeding depression for boll weight of Gossypium hirsutum.

50 40 30 20 Percentage (%) 10 0 -10 -20 -30 -40 -50 CIM-109 (P1) CIM-109 x CIM-240 CIM-109 x CIM-1100 CIM-109 x FH-682 CIM-109 x BH-36 CIM-109 x CRIS-9 CIM-240 (P2) CIM-240 x CIM-109 CIM-240 x CIM-1100 CIM-240 x FH-682 CIM-240 x BH-36 CIM-240 x CRIS-9 CIM-1100 (P3) CIM-1100 x CIM-109 CIM-1100 x CIM-240 CIM-1100 x FH-682 CIM-1100 x BH-36 CIM-1100 x CRIS-9 FH-682 (P4) FH-682 x CIM-109 FH-682 x CIM-240 FH-682 x CIM-1100 FH-682 x BH-36 FH-682 x CRIS-9 BH-36 (P5) BH-36 x CIM-109 BH-36 x CIM-240 BH-36 x CIM-1100 BH-36 x FH-682 BH-36 x CRIS-9 CRIS-9 (P6) CRIS-9 x CIM-109 CRIS-9 x CIM-240 CRIS-9 x CIM-1100 CRIS-9 x FH-682 CRIS-9 x BH-36

5 Boll weight (g) 4 3 2 1 0 -1 -2 -3 -4 -5

Cross combinations and Parental cultivars F1 Heterosis Inbreed depression (observed) F1 Means F2 Means Parents Means

Bolls per plant Bolls per plant varied from 24 (CIM-240) to 38 (CIM-109) among the parents and from 20 (CIM-1100 x CIM-240) to 53 (CRIS-9 x CIM-1100) among the F1 hybrids (Fig. 2). The maximum number of bolls per plant were recorded in five F1 hybrids of CIM-1100 as male or female parent viz; CRIS-9 x CIM1100 (53), CIM-1100 x CIM-109 (53) and its reciprocal (52), BH-36 x CIM-1100 (52) and CIM-1100 x CRIS-9 (51). It was closely followed by two other hybrids of CIM-1100 i.e. CIM-1100 x FH-682 (47) and CIM-240 x CIM-1100 (45). The cultivar CIM-240 (24) and its crosses with CIM-1100 (20), BH-36 (22), FH-682 (23) and its reciprocal (26) and CIM-109 (27), produced the least number of bolls per plant. In the F2 generation test, bolls per plant varied from 25 (CIM-240 x FH-682) to 44 (CIM1100 x FH-682) among the F2 population (Fig. 2). Maximum number of bolls per plant (44) were

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton

535

recorded in the F2 genotype CIM-1100 x FH-682 and it was followed by six other derivatives of CIM1100 as male parent viz; FH-682 x CIM-1100 (40), CRIS-9 x CIM-1100 (39), CIM-109 x CIM-1100 (37) and as maternal parent CIM-1100 x CRIS-9 (39), CIM-1100 x 109 (38) and CIM-1100 x BH-36 (38). The least number of bolls per plant (25) were produced by the F2 genotypes CIM-240 x FH-682 and BH-36 x FH-682. In F1 generation, where heterosis for bolls per plant was exhibited it ranged from +3.13 (FH682 x CRIS-9) to +65.63% (CRIS-9 x CIM-1100) over better parent (Fig. 2). Forty seven percent of the F1 hybrids attained positive heterosis. Nine crosses involving the cultivar CIM-1100 as a paternal or maternal parent manifested maximum and highly significant heterosis for bolls per plant ranged from +17.65 to +65.63% over better parent. In the remaining 21 crosses, two hybrids expressed significant heterosis over better parent. The majority of the F2 genotypes displayed inbreeding depression % for bolls per plant and the observed inbreeding depression ranged -2.78 to -32.69%. Some F2 population pose better mean values and less inbreeding depression % by having increased bolls per plant in F2 generation due to transgressive segregation.
Figure 2: Mean performance, heterosis and inbreeding depression for bolls per plant of Gossypium hirsutum.
100 80 60 40 Percentage (%) 20 0 -20 -40 -60 -80 -100 CIM-109 (P1) CIM-109 x CIM-240 CIM-109 x CIM-1100 CIM-109 x FH-682 CIM-109 x BH-36 CIM-109 x CRIS-9 CIM-240 (P2) CIM-240 x CIM-109 CIM-240 x CIM-1100 CIM-240 x FH-682 CIM-240 x BH-36 CIM-240 x CRIS-9 CIM-1100 (P3) CIM-1100 x CIM-109 CIM-1100 x CIM-240 CIM-1100 x FH-682 CIM-1100 x BH-36 CIM-1100 x CRIS-9 FH-682 (P4) FH-682 x CIM-109 FH-682 x CIM-240 FH-682 x CIM-1100 FH-682 x BH-36 FH-682 x CRIS-9 BH-36 (P5) BH-36 x CIM-109 BH-36 x CIM-240 BH-36 x CIM-1100 BH-36 x FH-682 BH-36 x CRIS-9 CRIS-9 (P6) CRIS-9 x CIM-109 CRIS-9 x CIM-240 CRIS-9 x CIM-1100 CRIS-9 x FH-682 CRIS-9 x BH-36 50 40 30 20 10 0 -10 -20 -30 -40 -50 Bolls per plant (No.)

Cross combinations and Parental cultivars F1 Heterosis Inbreed depression (observed) F1 Means F2 Means Parents Means

Seed cotton yield per plant Seed cotton yield per plant ranged from 67.85 g (CIM-240) to 86.62 g (CIM-109) among the parents and from 59.12 g (BH-36 x CIM-240) to 188.81 g (CRIS-9 x CIM-1100) among the F1 hybrids (Fig. 3). The highest seed cotton yield was recorded in three F1 hybrids CRIS-9 x CIM-1100 (188.81 g), BH36 x CIM-1100 (183.58 g) and CIM-1100 x CIM-109 (180.92 g). These were followed by the second top scoring hybrids i.e. CIM-1100 x CRIS-9 (169.98 g), CIM-1100 x FH-682 (164.16 g), CIM-109 x CIM-1100 (162.68 g), CIM-240 x CIM-1100 (160.56 g) and FH-682 x CIM-1100 (155.19 g). The parental cultivars of these promising crosses were having nearby half yield in comparison to their hybrids ranging from 67.85 to 86.62 g per plant. In F2 generation the seed cotton yield per plant varied from 63.63 g (CIM-240 x FH-682) to 138.10 g (CIM-1100 x FH-682) among the cross population (Fig. 3). The highest seed cotton yield (138.10 g) was obtained from the F2 genotype CIM-1100 x FH-682. It was followed by three other derivatives of CIM-1100 i.e. FH-682 x CIM-1100 (122.60 g), CIM-1100 x CRIS-9 (121.84 g) and its reciprocal CRIS-9 x CIM-1100 (119.96 g). Two other genotypes of CIM1100 also gained the third position in seed cotton ranging from 102.53 to 109.42 g. The above

536

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

derivatives of CIM-1100 surpassed all the hybrids and parents in yield. The lowest seed cotton yield per plant was recorded in the genotype CIM-240 x FH-682 (63.63 g). The CIM-1100 population with the maximum yield could be utilized in the segregating generations to evolve cultivars with better yield potential. In F1 generation (Fig. 3), 20 F1 hybrids had positive heterosis for seed cotton yield per plant and the increase ranged from +0.82 (CIM-109 x BH-36) to +115.22% (CRIS-9 x CIM-1100). Regarding the superiority over the better parent, 15 hybrids exhibited highly significant (P<0.01) heterosis, while one hybrid surpassed their better parent at the level of P<0.05. The crosses of the cultivar CIM-1100 with all the other cultivars (except as a maternal parent with CIM-240) manifested maximum and highly significant heterosis over better parent (+55.75 to +115.22%). All the F2 genotypes showed inbreeding depression for seed cotton yield per plant and the observed inbreeding depression ranged from -1.87 to -44.15%. In F2 generation, the derivatives of the cultivar CIM-1100 maintained stability for seed cotton yield (Fig. 3) even after segregation and inbreeding depression. The results in F2 generation provide good ground for further study in segregating generations.
Figure 3: Mean performance, heterosis and inbreeding depression for seed cotton yield of Gossypium hirsutum
120 100 80 60 40 Percentage (%) 20 0 -20 -40 -60 -80 -100 -120 CIM-109 (P1) CIM-109 x CIM-240 CIM-109 x CIM-1100 CIM-109 x FH-682 CIM-109 x BH-36 CIM-109 x CRIS-9 CIM-240 (P2) CIM-240 x CIM-109 CIM-240 x CIM-1100 CIM-240 x FH-682 CIM-240 x BH-36 CIM-240 x CRIS-9 CIM-1100 (P3) CIM-1100 x CIM-109 CIM-1100 x CIM-240 CIM-1100 x FH-682 CIM-1100 x BH-36 CIM-1100 x CRIS-9 FH-682 (P4) FH-682 x CIM-109 FH-682 x CIM-240 FH-682 x CIM-1100 FH-682 x BH-36 FH-682 x CRIS-9 BH-36 (P5) BH-36 x CIM-109 BH-36 x CIM-240 BH-36 x CIM-1100 BH-36 x FH-682 BH-36 x CRIS-9 CRIS-9 (P6) CRIS-9 x CIM-109 CRIS-9 x CIM-240 CRIS-9 x CIM-1100 CRIS-9 x FH-682 CRIS-9 x BH-36 180 150 120 90 60 30 0 -30 -60 -90 -120 -150 -180 Seed Cotton Yield (g)

Cross combinations and Parental cultivars F1 Heterosis Inbreed depression (observed) F1 Means F2 Means Parents Means

Lint percentage The lint % ranged from 31.83% (FH-682) to 36.44% (CIM-1100) among the parents and from 32.84% (CIM-109 x FH-682) to 36.92% (BH-36 x CIM-1100) among the F1 hybrids (Fig. 4). Maximum and statistically equal values of lint percentage were recorded in two derivatives of CIM-1100 in direct (36.92%) and reciprocal cross (36.79%) with BH-36 in F1 hybrids. Statistically equal values of 10 other hybrids (having 4 derivatives of CIM-1100), and 2 parents followed them. These values ranged from 35.58 to 36.61%. The minimum values of lint percentage enunciated by 3 parents (CIM-109, BH36 & CRIS-9) and 7 hybrids ranged from 32.84 to 34.06%. The lowest lint % was exhibited by cultivar FH-682 (31.83%). In F2 generation, the lint % ranged from 32.36% (FH-682 x CRIS-9) to 36.14% (CRIS-9 x CIM-1100) among the cross population (Fig. 4). CIM-1100 and its three crosses with CRIS9, CIM-240 and BH-36 ranging from 35.47 to 36.38% exhibited maximum and statistically equal mean lint %. For lint % in F1 generation (Fig. 4) 12 crosses expressed positive heterosis over their better parental values and the degree of increase ranged from +0.27 (CIM-109 x CIM-1100) to +3.88% (FH-

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton

537

682 x BH-36) and statistically, only one cross manifested highly significant heterosis (3.88%). Almost all the F2 population displayed inbreeding depression % for lint % and the observed inbreeding depression was -0.19 to -5.94%.
Figure 4: Mean performance, heterosis and inbreeding depression for lint percentage of Gossypium hirsutum

50

50 40 Lint perc entage (%

30

30 20

P ercentage (%

10

10 0

-10

-10 -20

-30

-30 -40

-50 CIM -109 (P 1) C IM -109 x CIM -240 CIM -109 x C IM -1100 CIM -109 x FH-682 CIM -109 x B H-36 C IM -109 x CRIS -9 CIM -240 (P 2) CIM -240 x CIM -109 C IM -240 x CIM -1100 CIM -240 x F H-682 CIM -240 x B H-36 CIM -240 x CRIS -9 CIM -1100 (P 3) CIM -1100 x C IM -109 CIM -1100 x CIM -240 CIM -1100 x FH-682 CIM -1100 x B H-36 CIM -1100 x CR IS -9 F H-682 (P 4) F H-682 x CIM -109 FH-682 x CIM -240 FH -682 x CIM -1100 F H-682 x B H-36 FH-682 x CRIS -9 B H-36 (P 5) B H-36 x CIM -109 B H -36 x CIM -240 B H-36 x C IM -1100 B H-36 x FH-682 B H-36 x CRIS -9 CRIS -9 (P 6) CRIS -9 x CIM -109 CRIS -9 x CIM -240 CR IS -9 x CIM -1100 C RIS -9 x F H-682 CRIS -9 x B H-36

-50

C ros s c om binations and P arental c ultivars F 1 M eans F 2 M eans P arents M eans

F1 Heterosis

Inbreed depres sion (obs erved)

Staple length In parental cultivars the staple length ranged from 25.66 mm (CRIS-9) to 29.10 mm (CIM-1100) among the parents and from 24.97 mm (CRIS-9 x BH-36) to 30.14 mm (CIM-1100 x FH-682) among the hybrids in F1 generation (Fig. 5). Maximum staple length (30.14 mm) was attained by the F1 cross combination CIM-1100 x FH-682, which was not significantly different from the hybrid CIM-1100 x CIM-240 (29.10 mm). In turn the second top performing cross was not significantly different from the parent cultivar CIM-1100, its 7 derivatives and two other hybrids ranging from 28.11 to 29.04 mm in fibre length. The cultivar CRIS-9 (25.66 mm) and its cross with BH-36 (25.81 mm) had the shortest fibers. In F2 generation the staple length varied from 25.47 mm (BH-36 x CRIS-9) to 29.84 mm (CIM1100 x FH-682) among the cross population (Fig. 5). In F2 genotypes the maximum staple length (29.84 mm) was also achieved by the same cross combination CIM-1100 x FH-682 as mentioned in F1 generation. It was not significantly different from the parental cultivar CIM-1100 (29.10 mm) and its 3 other derivatives [CIM-109 x CIM-1100 (28.96 mm) and its reciprocal (28.39 mm) and FH-682 x CIM-1100 (28.62 mm)]. The genotype BH-36 x CRIS-9 exhibited the lowest staple length (i.e. 25.47 mm). In F1 generation the 12 hybrids showed positive heterosis (Fig. 5) and the increase for staple length was ranged from +0.28 (CIM-240 x CIM-1100) to +6.00% (CRIS-9 x CIM-240) over the better parent. Two crosses attained maximum and significant heterosis for staple length over better parent (+4.65 to +6.00%). Most of the crosses having CIM-1100 as one of the parent displayed the highest range of heterosis for staple length. All the F2 population depicted inbreeding depression for staple length and the observed inbreeding depression ranged from -0.54 to -5.23%.
Figure 5: Mean performance, heterosis and inbreeding depression for staple length of Gossypium hirsutum

538

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

30 30 20 P ercentage (% 10 0 -10 -10 -20 -30 -30 -50 CIM -109 (P 1) C IM -109 x CIM -240 CIM -109 x C IM -1100 CIM -109 x FH-682 CIM -109 x B H-36 C IM -109 x CRIS -9 CIM -240 (P 2) CIM -240 x CIM -109 C IM -240 x CIM -1100 CIM -240 x F H-682 CIM -240 x B H-36 CIM -240 x CRIS -9 CIM -1100 (P 3) CIM -1100 x C IM -109 CIM -1100 x CIM -240 CIM -1100 x FH-682 CIM -1100 x B H-36 CIM -1100 x CR IS -9 F H-682 (P 4) F H-682 x CIM -109 FH-682 x CIM -240 FH -682 x CIM -1100 F H-682 x B H-36 FH-682 x CRIS -9 B H-36 (P 5) B H-36 x CIM -109 B H -36 x CIM -240 B H-36 x C IM -1100 B H-36 x FH-682 B H-36 x CRIS -9 CRIS -9 (P 6) CRIS -9 x CIM -109 CRIS -9 x CIM -240 CR IS -9 x CIM -1100 C RIS -9 x F H-682 CRIS -9 x B H-36 -40

10

C ros s c om binations and P arental c ultivars F 1 M eans F 2 M eans P arents M eans

F1 Heterosis

Inbreed depres sion (obs erved)

Performance of F3, F4 and F5 cross families F3 Selected cross families On the basis of CLCV, general performance and observation, the families of four crosses i.e. CIM-109 x CIM-1100, CIM-240 x CIM-1100, CIM-1100 x CIM-109 and CIM-1100 x CIM-240 were maintained and evaluated in the presence of standard cultivar CIM-446 (Table 2). The data were recorded on bolls per plant, seed cotton yield per plant, lint % and staple length. On average, the four crosses having 11 different single plant families superseded the standard cultivar (CIM-4466) for bolls per plant by 14.29 to 60.00 % and by 18.00 to 65.00 % for seed cotton yield per plant. In the case of fibre quality traits, all the cross families, except one, showed an increase for staple length that ranged from 5.06 to 14.01%. In the case of lint % only two families manifested an increase over the standard cultivar (0.27 to 0.82%), two exhibited the same values with the standard and the remaining 7 families showed decrease in lint % (Table 2). Some of the families of these crosses exhibited extra higher mean values as compared to the standard cultivar, which were could not be ignored and fruitful results were expected in the coming generations. So on the basis of their best performance, selection was applied in these four cross families and forwarded to F4 generation. F4 Selected cross families Out of 25 cross families originating from four crosses, on average, 17 families excelled the standard cultivar for bolls per plant and the increase ranged from 3.23 to 70.97%, 23 families with range of 13.85 to 112.31% increase for seed cotton yield per plant, three families with increase of 1.58 to 2.11% for lint % and 25 cross families with the range of 4.37 to 17.86% increase for staple length (Table 3). Some of the progenies of these crosses had extra higher mean values in comparison to standard cultivar and therefore fruitful results were expected in the coming generations. On the basis of their best performance, selection was applied in the promising families of these four crosses and forwarded to F5 generation.

S taple Length (m m

50

40

539

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

F5 Selected cross families Out of nine cross families, on the average, eight cross families excelled the standard cultivar (CIM-4466) for bolls per plant with increase of 26.92 to 107.69%, eight cross families with range of 27.59 to 112.07% increase for seed cotton yield per plant, six with the percent increase of 0.29 to 6.63% for lint % and all the nine cross families with the range of 0.78 to 7.39% increase for staple length (Table 4). Some of the single plant progenies had extremely high values for bolls per plant (58), seed cotton yield per plant (134 g), lint % (39.4%), staple length (28.5 mm) and which should be further studied. On the basis of their best performance, selection was made in the families of these crosses and forwarded to F6 generation.

Table 2: Performance of selected F3 families compared with the standard cultivar of Gossypium hirsutum.

Crosses Min 35 39 44 32 51 51 40 32 32 31 32 30 % +/+40.00 +54.29 +40.00 +14.29 +51.43 +48.57 +60.00 +31.43 +22.83 +45.71 +34.29 Min. 33.0 33.2 34.1 35.8 34.4 35.1 34.6 33.6 31.8 32.9 32.9 35.2 % +/-6.53 +0.27 -3.27 +0.82 -3.27 0.00 -0.27 -1.90 -3.81 0.00 -2.45 -

Fami-lies

CIM-109 x CIM-1100 CIM-240 x CIM-1100 CIM-1100 x CIM-240 CIM-1100 x CIM-109 Standard (CIM-446)

1 2 1 2 3 4 5 1 2 3 1 -

Bolls per plant Max. Mean 56 49 61 54 54 49 50 40 57 53 55 52 59 56 65 46 46 43 60 51 56 47 45 35

Seed cotton yield per plant (g) Min Max. Mean % +/100 152 125 +25.00 103 155 140 +40.00 131 160 146 +46.00 90 150 118 +18.00 151 170 158 +58.00 152 168 165 +65.00 105 155 140 +40.00 90 186 131 +31.00 93 160 123 +23.00 90 160 138 +38.00 88 155 131 +31.00 85 135 100 -

Lint % Max. Mean 36.6 34.3 39.3 36.8 36.8 35.5 38.8 37.0 36.5 35.5 38.3 36.7 38.5 36.6 38.2 36.0 38.6 35.3 42.2 36.7 38.7 35.8 38.5 36.7

Min. 28.5 26.4 26.9 24.4 26.4 27.4 25.9 24.9 26.7 27.4 26.9 24.4

Staple length (mm) Max. Mean % +/30.0 29.3 +14.01 29.0 27.7 +7.78 29.0 27.9 +8.56 26.9 25.5 -0.78 28.2 27.3 +6.23 28.2 27.7 +7.78 28.2 27.0 +5.06 27.9 27.0 +5.06 28.7 27.6 +7.39 30.0 28.3 +10.12 28.5 27.9 +8.56 26.2 25.7 -

Min.(Minimum values), Max.(Maximum values) % + / - [Percent increase (+) and decrease (-) of family means over standard cultivar (CIM-446)]

540

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

Table 3:

Performance of selected F4 families compared with the standard cultivar of Gossypium hirsutum.

Crosses Min 21 17 40 42 22 17 18 27 22 22 20 24 25 19 23 25 28 24 48 40 14 26 25 29 22 27 % +/0.00 +6.45 +67.74 +70.97 +6.45 -38.70 -32.26 -3.23 -19.35 +22.58 -32.26 -6.45 +6.45 +12.90 +3.23 -16.13 +41.93 +3.23 +70.97 +54.84 +16.13 +22.58 +48.39 +51.61 +6.45 Min. 32.0 32.6 36.5 35.8 38.1 37.2 33.8 33.2 34.7 33.8 36.0 33.6 34.6 35.9 32.2 34.2 33.7 32.7 32.1 35.6 35.3 33.1 35.5 36.2 34.0 37.8 % +/-6.05 -10.53 -2.63 -4.47 +1.58 +2.10 -8.16 -7.37 -2.37 -8.42 -1.05 -5.53 -6.05 +2.11 -7.37 -3.16 -5.79 -8.42 -13.42 -3.95 -3.68 -4.47 0.00 -2.37 -6.84 Min. 27.4 28.2 25.9 25.9 25.4 26.7 26.9 26.9 25.9 26.2 25.9 27.4 25.9 25.7 25.2 26.7 26.4 26.4 29.2 27.9 26.2 26.7 24.1 26.4 27.7 24.4

Fami-lies

CIM-109 x CIM-1100 CIM-240 x CIM-1100 ` CIM-1100 x CIM-240 CIM-1100 x CIM-109 Standard (CIM-446)

1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 11 1 2 3 4 5 6 7 1 2 -

Bolls per plant Max. Mean 46 31 59 33 63 52 60 53 44 33 21 19 23 21 39 30 28 25 51 38 21 21 34 29 51 33 55 35 41 32 26 26 58 44 37 32 57 53 56 48 52 36 54 38 56 46 53 47 43 33 37 31

Seed cotton yield per plant (g) Min Max. Mean % +/55 120 94 +44.62 44 154 85 +30.77 105 165 135 +107.69 111 158 138 +112.31 58 114 87 +33.85 50 62 55 +15.38 54 68 63 -3.08 79 117 89 +36.92 65 84 74 +13.85 65 152 111 +70.76 61 62 62 -4.62 72 101 85 +30.77 75 151 98 +50.77 57 164 103 +58.46 68 122 95 +46.15 73 76 79 +21.53 80 167 127 +95.38 68 108 92 +41.54 138 164 151 +132.30 114 162 138 +112.31 40 150 104 +60.00 75 155 108 +66.15 73 160 132 +103.08 80 150 129 +98.46 60 120 90 +38.46 57 69 65 -

Lint % Max. Mean 38.3 35.7 35.4 34.0 37.5 37.0 37.3 36.3 39.7 38.6 41.2 38.8 36.0 34.9 36.2 35.2 38.3 37.1 36.1 34.8 39.2 37.6 38.2 35.9 36.9 35.7 40.3 38.8 38.9 35.2 39.4 36.8 38.6 35.8 36.7 34.8 33.6 32.9 37.3 36.5 37.5 36.6 37.7 36.3 40.2 38.0 38.0 37.1 36.8 35.4 38.2 38.0

Staple length (mm) Max. Mean % +/29.0 28.3 +12.30 29.0 28.7 +13.89 26.9 26.4 +4.76 27.4 26.7 +5.95 29.0 27.2 +7.94 27.7 27.1 +7.54 27.7 27.4 +8.73 28.2 27.6 +9.52 27.7 26.9 +6.75 27.9 27.3 +8.33 26.7 26.3 +4.37 27.7 27.6 +9.52 27.9 26.9 +6.75 27.7 26.7 +5.95 28.2 26.9 +6.75 27.4 27.1 +7.54 27.4 27.1 +7.54 27.7 27.2 +7.94 30.2 29.7 +17.86 29.7 28.8 +14.29 26.9 26.7 +5.95 28.2 27.3 +8.33 27.4 26.9 +6.75 27.9 27.2 +7.94 28.2 28.0 +11.11 25.9 25.2 -

Min.(Minimum values), Max.(Maximum values) % + / - [Percent increase (+) and decrease (-) of family means over standard cultivar (CIM-446)]

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton 541

Table 4:

Performance of selected F5 families compared with the standard cultivar of Gossypium hirsutum.

Crosses % +/+26.92 +26.92 +34.62 +88.46 +107.69 +88.46 -15.38 +26.92 +57.69 Min. 34.0 33.8 32.3 34.6 33.9 33.7 35.0 34.5 34.3 34.6 % +/+1.44 +0.58 -2.59 +0.86 -0.29 +4.90 +2.02 +6.63 +0.29 Min. 26.9 25.7 26.9 26.7 25.9 26.4 27.2 25.4 26.8 24.4

Fami-lies

CIM-109 x CIM-1100 CIM-240 x CIM-1100 CIM-1100 x CIM-240 CIM-1100 x CIM-109 Standard (CIM-446)

1 2 1 2 3 1 2 1 2 -

Min 22 30 30 44 49 33 20 22 37 23

Bolls per plant Max. Mean 49 33 35 33 41 35 55 49 58 54 57 49 26 22 45 33 45 41 28 26

Seed cotton yield per plant (g) Min Max. Mean % +/50 112 74 +27.59 70 81 76 +31.03 70 95 82 +41.38 100 126 112 +93.10 112 134 123 +112.07 75 130 113 +94.83 47 59 51 -12.07 50 103 76 +31.03 85 105 95 +63.79 52 64 58 -

Lint % Max. Mean 36.7 35.2 35.7 34.9 35.4 33.8 35.7 35.0 35.2 34.6 38.7 36.4 35.6 35.4 39.4 37.0 35.3 34.8 34.8 34.7

Staple length (mm) Max. Mean % +/28.2 27.6 +7.39 28.5 27.1 +5.45 27.9 27.5 +7.00 27.2 27.0 +5.06 26.7 26.3 +2.33 27.4 26.9 +4.67 27.7 27.5 +7.00 26.9 25.9 +0.78 27.6 27.2 +5.84 26.2 25.7 -

Min.(Minimum values), Max.(Maximum values) % + / - [Percent increase (+) and decrease (-) of family means over standard cultivar (CIM-446)]

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton

542

Discussion
In F1 generation, the bolls per plant ranged from 20 to 53 among the F1 hybrids, while in F2, the bolls per plant ranged from 25 to 44. The seed cotton yield per plant ranged from 59.12 g to 188.81 g among the F1 hybrids, while in F2 population the range was 63.63 g to 138.10. In F2 generation the abrupt drop in yield and yield components was due to inbreeding depression and segregation. Soomro and Kalhoro (2000) also manifested that hybrids showing higher magnitude of heterosis was generally associated with higher amount of inbreeding depression suggesting dominant genes functioning for these traits. The results are in corroboration with the findings of Baloch et al. (1993a), Galanopoulou and Roupakias (1999) and Khan et al. (2000). The F1 and F2 derivatives of CIM-1100 exhibited highest heterosis and showed leading position due to maximum bolls per plant and increased seed cotton yield. Hassan et al. (1999) and Khan et al. (1999a) also mentioned that heterosis for seed cotton yield was associated with heterosis for boll number. F1 hybrids can be used for hybrid cotton production, which have been greatly supported by Arshad (1991), Salam (1991), Singh et al. (1993), Lakho et al. (2001), Solangi et al. (2001) and Solangi et al. (2002) as they also reported varying degree of heterosis over better parent for bolls and seed cotton yield per plant. The cultivars like CIM-240 and BH-36 and some of their crosses showed poor performance. The results in F2 generation were also encouraging and even after inbreeding depression, majority of the CIM-1100 F2 cross population displayed better mean performance as compared to their parents and the selection in these crosses provided transgressive gene recombination. Tang et al. (1993a) also revealed hat F2 population showed significant heterosis even over high parent. It is noteworthy that selection in the above cross families will be a great achievement for isolation and advancing promising lines for enhanced cotton production. F2 hybrids having extra ordinary performance could also be used as such for seed of F2 crop to increase the seed cotton yield on per unit area as it have been mentioned by Baloch et al. (1991), Baloch et al. (1993a), Baloch et al. (1993b) and Singh et al. (1993). In case of fibre quality traits, the lint % means values varied from 32.84% to 36.92% in F1 hybrids, while in F2 genotypes the range of lint % was 32.36% to 36.14%, while for staple length the said range was 24.97 mm to 30.14 mm in F1 hybrids and 25.47 mm to 29.84 mm in F2 generation. The decreasing level observed in F2 generation was due to inbreeding and segregation as also noticed by Baloch et al. (1993a) and Dever and Gannaway (1992) that F1 hybrids had longer fibre and F2 generation did not consistently produce longer fibre than parents. In F1 and F2 generations, most of the hybrids predicted negative heterosis over better parent except two hybrids for lint %, hence lint % and staple length manifested the low level of significance. Soomro and Kalhoro (2000) also noted small amount of heterosis. Lower magnitude of inbreeding depression for lint % and staple length indicated that additive genes were responsible for the expression of these traits. These results are also in line with the findings of Tang et al. (1993b) and Khan et al. (1999b) that very few F2 hybrids were superior in fibre properties to their high parent. Davis (1978), Davis (1979), Galanopoulou and Roupakias (1999) and Khan et al. (1999b) also reported that G. hirsutum x G. hirsutum hybrids were quite stable and did not show appreciable heterosis and the crosses appeared to be having a little potential for further improvement. In present studies some promising F2 genotypes showed increase in mean values of lint % and staple length as compared to better parent and it seems to be due to transgressive recombination which exhibited positive and some of them showed significant increase for lint % and selection in these crosses provided better chances for improvement in segregating generations. The fibres with magnificent lint % and staple length are greatly desired by textile industry for more lint production and for having fine and strong yarn with less twist. The fibers with shorter staple length badly affect the efficiency of the textile mill. Our inferences are also supported by Percy and Turcotte (1992), Tang et al. (1993a), Tang et al. (1993b), Galanopoulou and Roupakias (1999), Hassan et al. (1999), Lakho et al. (2001), Solangi et al. (2001), Dayo et al. (2002) and Solangi et al. (2002) and reported varying degree of heterosis for lint % and staple length in F1 and F2 generations. The results in advanced generations (F3, F4 and F5) revealed that majority of the cross families surpassed the standard cultivar (CIM-446) for bolls per plant, seed cotton yield per plant and staple

543

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman

length, while for lint % some cross families showed increased percentage over standard cultivar as also perceived by Dayo et al. (2002) that all the hybrid families excelled the check cultivar while some hybrids ginned more and measured longer staple than better check cultivar. These plant families are greatly desirable from yield and textile point of view and can result in the cultivars having yield potential with desirable fibre quality traits. The results are also in accordance with the findings of Bahadur et al. (1993) and Dever and Gannaway (1992) who reported maximum percent increase of yield, yield components and fibre quality traits through transgressive segregation against check, parental cultivars and F1 hybrids. The results also got support from the research conclusions of Tang et al. (1993a), Tang et al. (1993b), Galanopoulou and Roupakias (1999) and Kalhoro et al. (2002) who reported the same type of improvement in different cross families in segregating generations.

Conclusion
It is concluded that highest yielding F1 hybrids yielded progressively less in subsequent generations due to inbreeding depression and in advanced generations surpass the best cultivar, whereas some lowest yielding hybrids was stabilized due to low inbreeding depression. It is suggested that yield of the F1 did not predict the yield of the bulks in the advanced generations and the combined performance of the hybrids in the F1 and F2 generation could be a good indicator to identify the most promising populations to be utilized either as F2 hybrids or as a source population for further selection in advanced generations.

Heterosis and Inbreeding Depression and Mean Performance in Segregation Generations in Upland Cotton

544

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] Arshad, M. 1991. Use of cytoplasmic male sterility and genetic male sterility in production of commercial cotton hybrids. The Pak. Cottons, 35(1):29-37. Bahadur, S., Khan, I.A and Mehdi, S.S. 1993. The estimates of heritability and genetic advances for various plant traits in segregating population of G. hirsutum L. I. Yield and related characters. Pak. J. Agri. Sci., 30(3):280-283. Baloch, M.J., Lakho, A.R and Soomro, A.H. 1993a. Heterosis in interspecific cotton hybrids. Pak. J. Bot., 25(1):13 20. Baloch, M.J., Tunio, G.H and Lakho, A.R. 1993b. Performance of F1 hybrids from intrahirsutum crosses of Upland cotton. Pak. J. Sci. Ind. Res., 36(1):38-40. Baloch, M.J., Tunio, G.H and Lakho, A.R. 1991. Expression of heterosis in F1 and its deterioration in intra-hirsutum F2 hybrids. Pakphyton, 3:95-106. Bricker, B. 1991. MSTATC: A microcomputer programme for design management and analysis of agronomic research experiments. Crop and Science department, MSU East Mancing MI 48824, USA. Cook, C.F. 1909. Suppressed and intensified characters in cotton hybrids. U.S. Dept. Agri. Bull., 1947:127 (Quoted by Memon, 1963). Davis, D.D. 1978. Hybrid cotton: specific problems and potentials. Adv. Agron. 30:129-157. Davis, D.D. 1979. Yield and fibre properties of selected F1 hybrids. Proc. Beltwide Cotton Prod. Res. Conf. Phoenix.Ariz,p.71-72. Natl. Cotton Council of America, Memphis,TN. Dayo, G.N., Soomro, A.R., Anjum, R., Bano, S and Kalwar, G.H. 2002. Estimation of heterosis in Cotton Leaf Curl Virus resistant hybrids. Indus J. Plant Sci.1(1):53-56. Dever, J.K and Gannaway, J.R. 1992. Relative fibre uniformity in between parent and F1 and F2 generations in cotton. Crop Sci., 32: 1402-1408. Fehr, W.R. 1987. Principles of cultivar development. Theory and technique. Macmillan Pub. Comp. Inc., New York. 115-119. Fonseca, S.M. 1965. Heterosis, heterobeltiosis, diallel analysis and gene action in crosses of Triticum aestivum L. Ph.D Thesis Purdue Univ. USA. Galanopoulou Sendouca, S and Roupakias, D. 1999. Performance of cotton F1 hybrids and its relation to the mean yield of advanced bulk generations. European J. Agron., 11(1): 53 62. Hallaner, A.R and Miranda, J.B. 1986. Quantitative genetics in maize breeding. The Lowa State Univ. Press U.S.A. Hassan, G., Mahmood, G., Khan, N.U and Razzaq, A. 1999. Combining ability and heterobeltiotic estimates in a diallel cross of cotton (G. hirsutum L.). Sarhad J. Agri., 15(6): 563-568. Kalhoro, A.D., Soomro, A.R., Anjum, R and Kalwar, G.H. 2002. Seed cotton yield, lint percentage and staple length of F3 glandless cotton as affected by Cotton Leaf Curl Virus. Indus J. Plant Sci., 1(1):73-75. Khan, N.U., Abro, H.K., Kumbhar, M.B., Hassan, G and Khan, M. 1999a. Exploitation of heterosis can combat Cotton Leaf Curl Virus (CLCV) incidence in cotton (G. hirsutum L.). The Pak. Cottons, 43(3&4):21-33. Khan, N.U., Abro, H.K., Hassan, G., Kumbhar, M.B. and Khan, M. 1999b. Study of Heterosis in Upland cotton-I. Quality traits. The Pak. Cottons, 43(3&4):53-64. Khan, N.U., Abro, H.K., Kumbhar, M.B., Hassan, G and Mahmood, G. 2000. Study of heterosis in Upland cotton-II. Morphology and yield traits. The Pak. Cottons, 44(1&2):13-23. Krishnaswami, R and Kothandaramon, R. 1977. Heterosis in inter-specific crosses of Gossypium. Ind. J. Genet. Pl. Breed., 36:40-45. Lakho, A.R., Bhutto, H., Chang, M.S., Solangi, M.Y., Kalwar, G.H and Baloch, A.H. 2001. Estimation of heterosis for yield and economic traits in cotton. (G. hirsutum L.). Sindh Bal. J. Pl. Sci., 3:26-30.

545 [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34]

Naqib Ullah Khan, Gul Hassan, Moula Bux Kumbhar, Sungtaeg Kang, Ikramullah Khan, Aisha Parveen and Umm-E-Aiman Percy, R. G and Turcotte, E.L. 1992. Inter-specific hybrid fibre characteristics of cotton altered by unconventional G. barbadense L. fibre genotypes. Crop Sci., 32:1437-1441. Salam, C.A. 1991. Status of Intra-specific commercial hybrids in the Punjab. The Pak. Cottons, 35(1): 5-17. Singh, T.H., Randhawa, L.S and Chahal, G.S. 1993. Heterosis and combining ability for harvest index and economic yield in Upland cotton (G. hirsutum L.). Crop Imp. Soc. India, 6869. Solangi, M.Y., Baloch, M.J., Bhutto, H., Lakho, A.R and Solangi, M.H. 2001. Hybrid vigour in inter-specific F1 hybrids of G. hirsutum x G. barbadense for some economic characters. Pak. J. Biol. Sci., 4:945-948. Solangi, M.Y., Baloch, M.J., Solangi, M.H., Chang, M.S and Lakho, A.R. 2002. Performance of intra-hirsutum F1 hybrids (G. hirsutum L.). Asian J. Plant Sci.1(2):126-127. Soomro, A.R., and A.D. Kalhoro. 2000. Hybrid vigour (F1) and inbreeding depression (F2) for some economic traits in crosses between glandless and glanded cotton. Pak. J. Biol. Sci.3(12):2013-2015. Soomro, B.A., Chang, M.A and Soomro, A.R. 1991. Status of inter-specific commercial hybrids in Sindh. The Pak. Cottons, 35(1):19-28. Tang, B., Jenkins, J.N., McCarty, J.C and Watson, C.E. 1993a. F2 hybrids of host plant germplasm and cotton cultivars .1. Heterosis and combining ability for lint yield and yield components. Crop Sci., 33(4):700 705. Tang, B., Jenkins, J.N., McCarty, J.C and Watson, C.E. 1993b. F2 hybrids of host plant germplasm and cotton cultivars .2. Heterosis and combining ability for fiber properties. Crop Sci., 33(4):706 710. Waller,G.B., Wilson, F.D and Martin, J.H. 1981. Influence of phenotype, season and time-ofday on nector production in cotton. Crop Sci. 21:507-511. Waller, G.D and Moffet, J.O. 1981. Pollination of hybrid cotton on the Texas high plains, nector production and honey bees visits. Proceeding of 1981 Beltwide Cotton Conf., Natl. Council of America, Memphis, Tennessee, U.S.A. Wynne, J.C., Emery, D.A and Rice, P.M. 1970. I. Combining ability estimates in Arachis hypogeae L. II. Field performance of F1 hybrids. Crop Sci., 10(6): 713-715.

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.546-560 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector
Filiz Giray Uludag University, Turkey E-mail: giray@uludag.edu.tr Abstract Telecommunications services as an infrastructure investment are a key component in order to establish other reel sector, to buy advanced technology and to support manufacturing industries (Fenna, 1998:54). Information is increasingly as a fundamental factor of production. The driving force of phenomenon that is known as the new economy is digital and information focused telecommunications technology (Slovenia-First, 2000). New technological advances have redefined telecommunications services and have increased their importance. Telecommunications sector has a dynamic nature within the global economy. With rapid growth in technological advance and changes in Keynesian macro-economic policies, most countries had pushed to start reform in this sector. In the 1980s, many countries have started reforms toward this sector. The aim of this paper is a) to analyze Turkish Telecommunication Sector from the perspectives of its performance, b) to compare the Turkish and EU Telecommunication Sectors with regard to process of reform of telecommunications. So this comparison is light on the address the real cause behind the level of Turkish Telecommunication Sector. Keywords: Telecommunications, privatization, market liberalization, regulation

Introduction
The telecommunications sector is one of the most important sectors in economy as a whole. Telecommunications sectors potential benefits are not only economic but also social and cultural. This paper organized as follows. First, the traditional features of telecommunications sector (supporting other industry branches, natural monopoly, cross-subsidization, zero marginal cost of production, externalities and sunk costs) are summarized. The globalization trend and the expansion of technological improvements have changed some features of telecommunications sector. This situation has leaded to change of form of government intervention. Also in this section, the process of restructuring in telecommunications sector is explained. Secondly, this paper presents privatization of Turk Telecom that is dominant operator in Turkey. It is discussed historical developments in Europe and Turkey from point view of the privatization, liberalization and regulation. At the mid-1980s, telecommunications sector in Turkey like EU member countries was inadequate both in quality and quantity. It is believed that the comparative analysis can be more benefit. In this study, cross-country comparison consists of not only relation exists between the number of line in given system and price changed for services, but also effectiveness of operation and maintenance, cost and productivity of human resources, management efficiency, and government policies on tariffs and financing (Saunders, Warford, Wellenius, 1994:59). In addition the privatization and liberalization of this sector in EU and

547

Filiz Giray

its candidate countries has been substantial for Turkey both as a candidate country and faced with huge competition for allocation resources with these countries in future. EU countries started reform in telecommunications area in the 1980s. The same process is valid for liberalization in these countries. Privatization has become the most widely used and the most important phase of the reform process. The origin of privatization is constituted the fact that British Telecom was privatized in 1984. Towards the end of 1990s, privatization in most EU member countries had been completed. As delay, Turk Telecom was privatized in 2005. Finally the conclusions from the data and observations and major barriers for implementation the reform of telecommunications in Turkey contained the last section. 1. The Telecommunications Sector and The process of Changes in Sector Telecommunications sector has some features: Generating Economic Growth Dividend; Telecommunications infrastructure is a crucial element of social overhead capital. The spread of telecommunications services reduces costs of interaction, expands market boundaries and enormously expand information flows (Waverman, Meschi and Fuss, 2007:1). Supporting Other Industry Branches; telecommunications sector can be expected to attract new resources of capital, management and technology. Telecommunications presents a huge field for the application in technology. In addition the new technology in telecommunication field offers more powerful and sophisticated means to convey and coordinate information of a financial, commercial and legal manufacturing (Arup, 1993:261). Natural Monopoly; telecommunications sector is referred to be a natural monopoly due to these reasons: A natural monopoly can arise from economies of scale and/or economies of scope (Noam, 1992:57). A single firm can take advantage of economies of scale and grow larger than its rivals. Because in economics of scale, the long-run average costs continues to decline. A larger firm has always lower costs. It has cost advantage that allows it to grow (Heagel, Cook, 1996:68 and Rosen, 2002:313). While the number of subscriber increases, the cost of every added customer will be less. Natural monopoly arises from economies of scope as well as economies of scale. Economies of scope are said that there is shared equipment or common facilities that make producing goods together less expensive than producing them separately (Telecommunications Economics and Policy Issue, 1997: 23). For example, improving the size of the telecommunications network will lead to larger rate of increase per time due to its technological features (Millward and Parker, 1983: 206). Natural monopoly distorts the conditions of perfect competition and creates inefficiency in allocation of resources and unfair distribution of income (Singer, 1972:71). Cross-Subsidization; Telecommunications services show a cross-subsidies feature. The profit of one service compensates the loss of other services. This phenomenon is often called cross-subsidies. Cross-subsidization is essentially used to transfer among regions/customers. Thanks to system of cross-subsidization, long distance telephone services are sponsored local calls and international calls are financed domestic services (Willem, 1998:19, Canoy, Bijl and Kemp, 2004:153-154). Zero Marginal Cost of Production; the telecommunications sector has zero marginal costs of production. The production of services in question may be increased without decreasing other production by using scarce resources apart from its production. In that event, price equals marginal cost, so that price is zero because marginal cost is zero (Herber, 1971: 32). Sunk Costs; telecommunications investment entails sunk costs. A firm included in telecommunications cant compensate costs from which it has to suffer when it has to leave out market in question. Therefore large sunk costs create a barrier to entry in market (Cave, 2004:34). Externalities; externality is likely to be important in the early years of establishment of a network where the value of a new subscriber may be anticipated to be highly valued by existing subscribers (Telecommunications Economics and Policy Issue, 1997: 53). Especially, telecommunications services advanced have externalities (Trotman-Dickenson, 1996:60 61). There are basically three ways of the social and economic benefits of network externalities. The first is to create interconnection among platforms. Namely network externalities arise from getting two side of a market together: sellers and buyers, users and software developers. There is substantial cross-subsidization between the two sides. The second way is to take stock of the existence of multiple, non-interconnected platforms. This third way creates scope for platform incompatibility (Tirole, 2004:262-263).

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector

548

Until the 1980s telecommunications operations in countries (the exceptions were Canada, Finland and The United States) had been public enterprises. However, these state monopolies generally fell short of meeting needs. In the 1970s, in order to solve this problem, countries focused on obtaining a larger share of their government budgets and external credit and aid for telecommunications, but public enterprises didnt provide these services neither in good quality nor in enough quantity. Since the mid 1980s, governments turned to microeconomic reforms in an attempt to improve the competitiveness of their economies. Economic policy shifted towards a greater role for market forces and a disengagement from the economy by state. Telecommunications as an infrastructure is one of areas that are virtually in need of restructure. This is due to fourth main factors: First is that together with the globalization trend, the expansion of technological improvements and industrialization changes in the monopolistic characteristic of telecommunications sector. Technological change effects natural monopoly two ways: The first effect is the fact that the structure of cost in natural monopoly had been changed thanks to technological improvements. The second effect is result of abolishing natural monopoly feature of infrastructure investment. Sector obtains a competitive structure and new goods and services that can be substituted for monopoly goods are created. These two effects create important changes on telecommunications and electricity sector. The one of most important effects of technological improvement in telecommunications sector declines network cost. The cost of optical fibber systems continues to fall even as capacity increases; the new format for transmission systems permits flexible and inexpensive access to data streams. All of the developments are reducing costs (Saunders, Warford, Wellenius, 1994: 325). Also, technological improvements increase competition in many sections of telecommunications networks. For example, increase number of firm that was included in telecommunications sector was raised from % 24 to % 34 in 1996 (Kessides, 2004:39). Technological, market and international developments have made the reshuffling of cross-subsidization (Willem, 1998:81). The low scale production occurred in some sectors thanks to change of structure of natural monopoly. Noam claimed that a second network enters in market and its rivalry leads to reduce of costs at each level of production. That is, economizes of scale doesnt dispose of the question of lower cost. Larger telecommunications networks offer advantages in the form of cost saving due to economies of scale. But larger telecommunications networks riskier because they influence more customers when it fails (Smith, 2006:299). The failure of a big resource can have a large impact. For instance, If the failure of some or all of the trunks leads to lost calls, It occurs the lost revenue based on three factors, the average rate at which calls are lost, the revenue per call, and the stage duration. It must be compared the rewards and risks of larger infrastructure (Smith, 2006:305). Economizes of scope have been changed. That is, two are more products are produced by one firm that can has level of different technology together with technological improvement. For example in communications services, while telecommunications services are the density of capital, postal services are the density of labour. For this reason, these services can be divided into two services. So, each one opens up competition. Three new economics theories towards competition in markets were constituted for opening up competition in natural monopoly. First theory is called Demsetz Competition Theory. If according to this theory, competition isnt possible in market, competition can be ensured to enter in market (Demsetz, 1968:55-56). The second theory is named Contestable Markets under the existing potential rivals who will join in market, one firm in market avoid being monopoly (Baumol, Panzer and Willing, 1982). The last theory was developed by Chamberlin is Chamberlinian Monopolistic Competition Theory (Chamberlin, 1933). Second: the second problem in telecommunications sector is the lack of effective competition. Third: Telecommunications services have fallen far short of needs. Fourth; nowadays, in the telecommunications sector, there are new value added services such as analogue mobile phone, Internet, satellite services, data transmission etc. in addition to the classical services.

549

Filiz Giray

Process reforming in telecommunications sector that has been need restructure changed form of intervention of government in this sector. Instead of supplying telecommunications services directly, the governments are provided these services by the private sector. For this reason all over the world, nearly all countries have preferred four processes to the reform toward telecommunications. These are commercialization, corporation, privatization and liberalization. More and more both developed and developing countries have privatized their state telecommunications enterprises. Although some countries attempted radical restructuring that would realize an independent, autonomous, government-owned commercially oriented under government control. The reforms didnt response to modernize the telecommunications sector. The timing, method and preparation are crucial to be successful. Under the reform of telecommunications sector, the question is not whether to reform, but rather how and when (Saunders, Warford, Wellenius, 1994:322, 328). After reform, telecommunications services must become more and more efficient and relatively less and less costly. 2. In Terms of the Privatization, Liberalization and Regulation a Comparison between the European and Turkish Telecommunications Sectors Both Turkish and European telecommunications sectors nearly showed the same progress until the end of 1970s. In Turkey, The post, Telegraph and Telephone (PTT) General Directorate was established as a state-owned monopoly in 1924. In 1994, PTT was split into two organizations as The Directorate of Postal Operations and Turk Telecom that is responsible for telecommunications services. Turk Telecom has still a monopoly over all basic telecommunications services. At the end of 1970s, Turkish Telecommunications Sector had serious problems that were similar to the EU countries. For example, the necessary investment funds for upgrading its network; the long waiting lists for telephone connections; over bureaucratic on PTT administrations, abuse of monopoly power due to political interference, storing of information without efficient and effective and not responding to the differentiated demands of both businesses and consumer groups. Sometimes the PTT that was used by both Turkish government and Europe countries governments to achieve other objectives than providing universal telecommunications services: competition policies, innovation policies, macro-economic policy, employment and regional policies (Willem, 1998:20-21). However Turkish Telecommunications Sector has had more comprehensive and serious problems than EU countries. In EU, telecommunications services were supplied by that central government through statutory monopolies until 1980s. In EU countries, the monopoly in the telecommunications area was organized in variety of way: functionally separated monopolies (i.e. between the operation of national and international services e.g. Italy), geographically subdivided monopolies (i.e. on a regional basis e.g. Denmark) and de facto monopolies (either statutory as in Belgium or exclusively operated by a semipublic corporation as in Spain). In most cases, the telephone services were operated by PTT (Willem, 1998:76). The increase of state intervention at the sector level was parallel by the establishment of the welfare state at the macro-level. At the beginning of 1980s, European economies were faced with economic recession and mounting competition the newly industrializing countries. In 1980s, national policy makers aimed to strengthen the key sector of countries and to prepare international competition. European states were forced to respond demands and their national market were too small to complete with the US and Japan (Willem, 1998:15-16, 26). The process of liberalization that had been started in the USA was followed by the UK and Japan. Telecommunications separated from postal services by the Thatcher government in 1981. British Telecom was established in corporate status in 1984. The privatization of British Telecom was realized to permit the sale of 50, 2 %, the rest of British Telecoms share was sold in 1991 and 1993. Other Europe countries have followed the UK for privatization in telecommunications area. In 1990s, most EU member countries completed their privatization process in telecommunications areas. After privatization, the productivity of British Telecom rose by an average of 7.5 percent between 1983 and 1992 (Healey and Cook, 1996:71).

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector

550

In Turkey, the history of privatization in telecommunications extends until the beginning of 1990s, but the pace of privatization had slowed down in comparison with EU member countries. The basic reasons of this case ranked below: A political and social consensus couldnt be made for privatization: This situation based on two main reasons, Firstly telecommunications is a strategic sector aspects security and defence. Other reason is the fact that the priority objectives of privatization were accepted to generate incomes for government through the sale of share and annual tax revenues. Political instability and uncertainty: Governments rapidly has changed. To face serious economic problems like high inflation rate, unemployment and heavy debt burdens. To meet constitutional cancellation: Laws on privatization enacted were often turned down from the Supreme Court. Lack of board to organize the program of privatization. When the Turkish Telecommunications Sector is evaluated from the point view of country risk, sector risk and company risk. It has great potential for privatization. Turk Telecom has huge capacity with new investment opportunities for domestic and foreign investors. We can see these potentials from some Turk Telecom indicators. Moreover, it has had a fast growing economy. First privatization law was enacted in May 2001. This law included many incomplete and faults. As a result, privatization couldnt be realized with this law. However, the privatization was unsuccessful not only due to this law but also political instability and poor market conditions. Finally, in November 2005, 55 % of shares of Turk Telecom had bought by international shareholders, the sale strategy based on block sale (Saudi Ojer Telecommunications Company and Telecom Italia Consortium). The second main component of telecommunications reform is liberalization. Liberalization of telecommunications sector doesnt simply mean the removal of barriers to market entry and the establishment of a level playing field between participants. Telecommunications sector requires a more interventionist stance, tipping the playing field by imposing asymmetric regulatory obligations upon incumbents to positively assist new entrants (Walden, 2005:129). The UK carried out liberalization and privatization together. In 1982, Mercury was given a twenty-five year license to run long-distance communications systems. In 1985, Mercury took a fully operational digital system. Therefore, the UK had duopoly shared between British Telecom and Mercury. Also the equipment market in the UK was opened up to competition (Willem, 1998:145). The market for telecommunications services in the EU was fully liberalized at a regulatory level at the end of 2000 (Walden, 2005:122). In the all of Europe, telecommunications had been liberalized under a traditional antitrust approach. The restructuring in EU resulted both in competition and in vertical integration with USA (Aufderheide, 1999:26). Therefore most European states privatized their telecoms. They have opened up their market for free competition. In 1987, Green Paper Provision liberalization-oriented was published by EU. Common goals are: (1) Liberalization of area under a monopoly provider, (2) Opening access to telecommunications networks and services, through harmonization and the development of minimum standards, (3) Full application of the competition rules (Walden, 2005:110). The European Council held in Lisbon on 23-24 March 2000 set the goal for Europe for the next decade to became the most competitive and dynamic knowledge-based economy in the world. In response to this need the eEurope Action Plan was accepted on the 19-20 June 2000. In Lisbon, It has been target the fact that telecommunications sector will be more dynamic and competitive sector by 2010 in the world (eEeurope + 2003 Progress Report, February, 2004: 8) The main objectives of Plan is:

551

Filiz Giray

(1) Accelerate the putting in place of the basic building blocks for the information Society, (2) A cheaper, faster, secure internet, (3) Investing in people and skills, (4) Stimulate the use of the internet (eEeurope + Action Plan, (June, 2004: 1-2). The telecommunications restructure process in fact opened up competition and lowered prices without wrecking the network. Even if the common belief of free market theorists is that the market will determine the optimal level of technology necessary without government regulation. Because telecommunications sector has different cost structure and revenue expectations with respect to reform of telecommunications, the regulatory authority is needed. Business serves customers where there is a profit to be made. According to this fact, telephone companies dont often do so in many rural and poor areas of the country. That is why, it is more expensive to lay telephone network in rural areas. Particularly advanced services can reduce costs, but it is too difficult to practice (Aufderheide, 1999:263). At the European level, several measures have been taken to promote and hasten the transition to information society. The most significant among these are the acceleration of the liberalization of telecommunications. First phase in order to achieve the goal of liberalization the telecommunications sector in EU was the 1998 legislative regulator package. This package consisted of control of retail prices, control of access prices and universal service obligation. Later the new regulator legislation in the EU was approved and published by the European Parliament and Council in 2002. The fundamental purposes of the new legislation are; to provide services like a common regulatory framework and principles of competition, to design operators enjoying significant market power or which ex ante obligations may be imposed, to provide completely convergence between member states (Buigues, 2004:9-10, 16). Competition is a key factor in lowing prices (Figure 1).
Figure 1: DSL (Digital Subscriber Line) Prices and Degree of Competition
12 0 10 0

* PT * ES * LU * FR * DE
0 0,1 0,2 0,3 0,4

D SL Prices

80 60 40 20

* IT

* UK * SE
0,5

* NL
0,6 0,7

D erg re e o f C o mpetitio n

Source:

EUROSTAT, Information Society Benchmarking Report, 2004.

The liberalization process of the Turkish Telecommunications Sector is as follows: Liberalization of telecommunications equipment sector was performed in 1993. Liberalization of value added telecommunications services, with mobile phone was partially performed in 1994. Initially, two private firms were given licensing for mobile phone through revenue sharing agreements. Later, the operators license was sold by means of a concession agreement. In year 1998, the government sold the two operators licenses. Full liberalization in this area hasnt been carried out yet. A successful privatization requires a good regulated framework. In 2000, finally a law was passed to establish a regulatory authority called Telecommunications Authority. Telecommunications authority determined licensing conditions such as applicants, time, limits and procedural in 2002. The different arrangements for licensing from 1994 to 2002 leaded to confuse and abuse. Licensing is a key aspect of telecommunications regulation. Thanks to licensing can control market entry, therefore, can

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector

552

be used to shape the market by limiting or not the number of players or the types of services (Flanagen, 2005:153). Turkish Telecommunications sector is far behind EU countries and its candidate countries aspect to both privatization and liberalization. This lag has caused to continue problems in telecommunications area and to decline competitive power with EU countries. This case in Turkish Telecommunications Sector is showed follow in the light of quantities indicators:
Figure 2: Annual Investments in Telecom (106)

Telecommunications Investment
6000 5000 4000 3000 2000 1000 0 Turkey
Source: ITU, Turkey and EU Indicators, 2005.

4781

1861

EU Average

Figure 3: Per Capita Telecommunications Investment (USA $)


Country England (1990) France (1990) Spain (1991) Turkey (1990)
Source: ITU, World Telecommunications Indicator, 1992.

Per Capita Telecommunications Investment 82.0 87.4 150.2 21.9

In Turkey, the gap between necessary and actual telecommunications investment had continuously increased due to the governments budgetary constraints. Unsatisfied telecommunications investment can be seen comparable with EU member and candidate countries in Figure 2 and 3. According to the ITU, at least 40 percent of telecoms revenues of countries should be transferred into investment (Deane and Opuku-Mensah, 1997: 10). There is a gap in favour of expenditures between telecoms revenue and telecoms expenditure. This gap stems from ineffective management, lacking of control and unbalanced tariff structure. The modernizing of networks requires considerable investment. Most EU and their candidate countries have completed the modernization, Turkey is behind them. However, it can be seen good progress the fact that Turkey has reached over % 90 digitization of its network (eEurope + 2003 Progress Report, February 2004:7). When compared to EU average, a teledensity rate (fixed line in service /total population) of 29 % in 2003 is low. EU average of this rate is approximately 35 %. The target rate for Turkey was 33 % until 200, but this target hasnt been reached yet (Europe + Final Progress Report, 2003). Fixed line penetration rate is lower than percentage of household having fixed line service. The structure of large families in Turkey is potentially effect on this result.

553
Figure 4: Fixed Lines Penetration Rates (%)
Number per 100 population

Filiz Giray

70 60 50 40 30 20 10 0 TR EU AVER. 2001 2002 2003 2005

Source:

eEurope + 2003 Progress Report, February, 2004 and ITU, Turkey and EU Indicators, 2005.

Figure 5: Percentage of Households That Have Fixed Telephone Service (%)


87

88 86 84 82 80 78 76 74 72

86

77

Turkey

EU Average

CC Average

Source:

ITU, World Telecommunication Indicator, March 2001

ITU statistics show that over the last 10 years, the fixed telephone lines have slowed down (ITU, World Telecommunication, 2006:4). Mobile penetration rate have continuously increased. Also, take up of mobile is increasing faster than fixed line take up. Namely, It is seen fixed-mobile substitution (Figure 6). The main reasons for mobile growth have been the introduction of prepaid services, rapid network development, and a higly competitive environment (ITU, Covering Note, 2006: 5).

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector
Figure 6: Penetration Rates (Fixed + Mobile)
160 140 120 100 80 60 40 20 0 1996 1999 2005 1995 1997 1998 2000 2001 2002 2003 Europe Turkey

554

Source:

eEurope + 2003 Progress Report, February, 2004 and ITU, Telecommunications Indicator, 2006, ITU, Turkey and EU Indicators, 2005.

Even If they are far behind EU countries. These indicates show a huge growth potential for mobile market from point view of Turkey. Some of the new member states rank quite high in terms of this comparison. Mobile penetration rate is closely related to the GDP per capita in each country (Central and Eastern, 2004:3). The international traffic inside and outside the country reflects another aspect of sector evaluation. The international traffic of Turk Telecom has increased.
Figure 7: International Incoming Telephone Traffic (106).

Minutes per subscriber

1800 1600 1400 1200 1000 800 600 400 200 0

2005

rk ey

Source:

ITU, Turkey and EU Indicators, 2005

The international traffic of Turk Telecom was average 33 % between 1995 and 1997. This rate reached 38.7 % in 2001. Increasing of this rate has showed without parallel growing population in Turkey.

Av er ag e

Tu

555
Figure 8: International Outgoing Telephone Traffic (106)

Filiz Giray

Minutes per subscriber

900 850 800 750 700 650 2005

Tu rk ey

Source:

ITU, Turkey and EU Indicators, 2005

Figure 9: Faults Per 100 Main Lines Per Year

60 50 40 30 20 10 0 Turkey EU Average CC Average

Source:

ITU, World Telecommunication Indicator, March 2001

The faulty rate that is one of the indicators of quality telecommunications services isnt very pleasant as seen Figure 9. The rate of Internet users in Turkey is low with compare to EU countries. The some basic reasons of this case: First is to be high of the price internet services; secondly, generally education level is low; third is related to shortage of internet facilities. The internet can provide a platform to enhancing countries social and economic development, that is opening up new market and revenue streams to business (ITU, World Telecom, 2006: 5). The highest penetration rates have been in Europe and Americas.

EU

ve ra ge

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector
Figure 10: Individuals Regularly Using the Internet (%)

556

45 40 35 30 25 20 15 10 5 0
Source: eEurope + 2003 Progress Report, February, 2004, Europe in Figures Eurostat Yearbook 2005, European communities, 2005 and Eurostat Indicators.

2001 2002 2003 2004 2005

Conclusion
At the end of the 1970, the restructuring in telecommunications sector for both developed and developing countries was a major goal. Changes in telecommunications policy were parallel by not being valid of Keynesian macro-economic policies. The free market economy requires becoming to prepare for international competition. Telecommunications services were provided by public enterprises as a monopoly for a long time. This paper presents the process of reform in Turkish Telecommunications Sector with compare to EU countries. After privatization and liberalization, EU countries have responded to the differentiated demands of both business and costumers group by competitor operators that are more flexible, less expansive. At beginning of 1980s, Turkey almost had the same problems as EU countries. Despite efforts, the pace of privatization has slowed down a comparison with other countries. The privatization of Turk Telecom was average fifteen years behind many countries practices. This lag has based on many reasons. For example, a political and social consensus couldnt be obtained for a long time. However, to some extent successful in privatization depends on a broad consensus. As a result of this lag, development in Turkish Telecommunications Sector is far behind EU member and candidate countries. It can be seen that this case based on indicators found by hitherto. Some statistics show that the gap between EU and Turkey hasnt been narrowing in terms of fixed telephone line, mobile subscriber, Internet users and investment. The performance of sector depends on to the extent to improve in these indicators. High growth rate in telecommunications area is to bring to Turkey risk falling behind in terms of Internet access and newer technology. Besides the recent, privatization law on telecommunications isnt satisfied expectations. For instance, the sale strategy was determined only block sale in law in question. The law doesnt clarify. Openness is important for a successful privatization policy. EU countries have preferred to use public offering. Although there are exceptions (e.g. Argentina, Mexico), the one of dominant factors that makes privatization successful is a good privatization law in telecommunications. The reform in telecommunications requires a process. The second main component of this reform is liberalization. In the field of telecommunications, the avoidance of monopoly rents requires the competition. By 1990s, operator in most of EU countries was opened to competition. In Turkey, although privatization was realized, liberalization hasnt been performed yet. Competitive telecommunications sector in EU is reaping the benefit of growth. The impact of liberalization of telecommunications services in EU and its candidate countries has been more important for Turkey. In Turkey, Turk Telecom can be transferred to private monopoly. The lack

557

Filiz Giray

of liberalization will hamper the sector. It will be face the same type of problems- high prices, low quantity services-. This situation, the competition power of Turkey will more and more reduce. Other issue concern to regulatory in Turkey. The subjects such as independent of the regulatory and licensing conditions, prices have some problems. The regulatory authority must be independent both financially and functionally. The regulation framework needs adapting to the EU new regulatory law. On the other words, if Turkey is to be competitive in the knowledge economy, it will have to adopt plans similar to those of the "Europe" action plan

The Privatization and Liberalization of Telecommunications: A Comparative Analysis for Turkish Telecommunication Sector

558

References
[1] [2] [3] [4] [5] [6] Arup, C. (1993), Innovation, Policy and Law: Australia and the International High Technology Economy, Cambridge, Cambridge University Press. Aufderheide, P. (1999), Communications Policy and the Public Interest, New York, the Guilford Press. Baumol, W. Panzar, J. and Willing, R. (1982), Contestable Markets and Theory of Industry Structure, New York. Buigues, P.A. (2004), the Competition Policy Approach, the Economics of Antitrust and Regulation in Telecommunications, (Ed: P.A. Buigues, P. Rey), Cheltenhem, Edwar Elgar Publishing Limited. Canoy, M. Bijl, P.D. and Kemp, R. (2004), Access to Telecommunications Networks, the Economics of Antitrust and Regulation in Telecommunications, (Ed: P.A. Buigues, P. Rey), Cheltenhem, Edwar Elgar Publishing Limited. Cave, M. (2004), Economic Aspects of the New Regulatory Regime for Electronic Communications Services, the Economics of Antitrust and Regulation in Telecommunications, (Ed: P.A. Buigues, P. Rey), Cheltenhem, Edwar Elgar Publishing Limited. Chamberlin, E.H. (1933), the Theory of Monopolistic Competition, Cambridge. Deane, J. and Opoku-Mensah, L. (April, 1997), Development and the Market: The Promises and the Problems, Telecommunications, (Ed: R. Lamb), Panos Media Briefing. Demsetz, H. (1968), Why Regulate Utilities, Journal of Law and Economics, 11. Elias, D. (April, 1978), The Possible Development of Telecommunications and Its Effects on the Telecommunications Industry, Philosophical Transactions of the Royal Society of London, 289, No:1356. eEurope + 2003 Progress Report (February, 2004). Europe + Final Progress Report (2003). Europe + Action Plan (June, 2004). Europe in Figures Eurostat Yearbook 2005 (2005), European Commission. Eurostat (September, 2004), Central and Eastern European Countries Information Society Benchmarks Summary Report. Eurostat (2004), Information Society Benchmarking Report. Fenna, A. (1998), Introduction to Australia Public Policy, Melbourne, Longman. Flanagan, A. (2005), Authorization and Licensing, Telecommunications Law and Regulation, (Ed: I. Walden and J. Angel), Oxford, Oxford University Press. Healey, N. and Cook, M. (1996), Supply Side Economics, (3th ed.), Guildford, Biddles Ltd. Herber, B. P. (1971), Modern Public Finance the Study of Public Sector Economics, Illinois, Richard D. Irwin Inc. Kesside, I. N. (2004), Reforming Infrastructure-Privatization, Regulation and Competition, Oxford, Oxford University Press. Millward, R. and Parker, D.M. (1993), Public and Public Enterprise:Comparative Behavior and Relative Efficiency, Public Sector Economics,New York, Longman Inc. Noam, E. (1992), Telecommonuciatons in Europe, Oxford, Oxford University Press. ITU (March, 2001), World Telecommunication Indicator. http://www.itu.int/TTUD/ict/publications/world/material/handbook.html. ITU (1992), World Telecommunications Indicator. ITU (2006), World Telecommunication /ICT Development Report. , Measuring ICT for Social and Economic Development. ITU (2006), Covering Note. ITU (2006), Telecommunications Indicator, htt:www.itu.int/ITU-D/ct/cs/letters/turkey.html. Rosen, H. S. (2002), Public Finance, (6th ed.), New York, McGraw-Hill/Irwin.

[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29]

559 [30] [31] [32] [33] [34] [35] [36] [37] [38] [39]

Filiz Giray Saunders, R. J., Warford, J.J., Wellenius, B.(1996), Telecommunications and Economic Development, (2nd ed.), Baltimore, Slovenia-First Among Eastern European Countries in Development of Telecommunications (2000). www.ukom.gov.s./eng/slovenia/background-information/telecommunications/ Singer, N. (1972), Public Microeconomics, Boston, Little Brown and Company. The John Hopkin University Press. Smith, E. D. (2006), How Big is Too Big? Trading off the Economies of Scale of Larger Telecommunications Network Elements Against the Risk of Larger Outages, European Journal of Operational Research, 173. Telecommunications Economics and Policy Issue (1997). Tirole, J. (2004), Telecommunications and Competition, The Economics of Antitrust and Regulation in Telecommunications, (Ed: P.A. Buigues, P. Rey), Cheltenhem, Edwar Elgar Publishing Limited. Trotman-Dickenson, D.T. (1996), Economics of the Public Sector, London, Macmillan Press Ltd. Walden, I. (2005), European Union Communications Law, Telecommunications Law and Regulation, (Ed: I. Walden and J. Angel), Oxford, Oxford University Press. Waverman, L. Meschi, M. And Fuss (2007), The Impact of Telecoms on Economic Growth in Developing Countries, www.ainic.info/workshop05/Waverman.ppt. Willem, H. (1998), Privatisation & Liberalization in European Telecommunications: Comparing Britain, the Netherlands & France.

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.560-573 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE
aban elik Yasar University Faculty of Economics and Administrative Sciences Business Administration Abstract The main indicators of risk for a particular asset or a portfolio are the standard deviation which we define it as an indicator of total risk and the beta which we define it as an indicator of systematic (market) risk. In this paper, the risk & return relationship of common stocks will be explained and analyzed for the purpose of constructing a portfolio in the context of Modern Portfolio Theory and Capital Asset Pricing Model. The main purposes of this research are (1) to measure the risk of common stocks and show their relations with the market portfolios return based on theoretical framework of Modern Portfolio Theory and Capital Asset Pricing Model, (2) to analyze a specific types of investment instruments (common stocks) from a specific sector (banking sector) for certain period of time (one year,2006). The distinctiveness of the study as a part of general research is to calculate systematic risk indicator, beta, for nine biggest banks common stocks and examine their risk & return relationship in context of Modern Portfolio Theory and Capital Asset Pricing Model based on daily return for 2006 in Turkey. Results indicated that (1) market risk indicator, beta, of Turkish Banks common stocks is much higher than those of USA, (2) there is not any negatively correlated common stocks in American Banks common stocks on the contrary of one negatively correlated common stocks in Turkish Banks common stocks, (3) volatility of American Banks common stocks are lower than those of Turkish, and (4) risk & return relationships is not totally supported by CAPM. Keywords: 1) Modern Portfolio Theory, 2) Capital Asset Pricing Model, 3) Risk Indicators 4) Banks Common Stocks JEL Classifications:

1. Introduction
There is no doubt that anything related to future which is supposed to be taken into consideration has to be predicted on the basis of risk. To comprehend risk we should look at two streams flowing through the 20th century. One is subjective probability. The other one is operationalism. Holton reviewed the literature and concluded that according to objective interpretations, probabilities are real. We may discover them by logic or estimate them through statistical analysis. According to subjective interpretations that probabilities are human beliefs. They are not intrinsic to nature. Individuals specify them to characterize their own uncertainty1 . Furthermore, a scientist has no chance to make a hundred
1

Holton,2004

561

aban elik

percent correct judgment about what is going to be in the future. In financial world, predictions, estimations, forecasting or even intuition are not wholly separated from probability which refers to the chance that event will occur. Moreover, there is no guarantee that those predictions are going to be correct. A scientist can not know what is going to happen whereas they are trying to explore at which level and under some assumptions what is supposed to be. Even though the definitions of risk vary among scientist, there are two obvious characteristics or/and components of risk: exposure and uncertainty. Risk, then, is exposure to proposition of which one is uncertain. Suppose a man leaps from an airplane without a parachute. If he is certain to die, He faces no risk. Risk requires both exposure and uncertainty2 . Harry Markowitz in his groundbreaking paper in financial literature of Portfolio Selection3 , showed that a trade-off between the expected return of the individual security and the contribution of that security to portfolio risk are determinant for portfolio optimization. William Sharpe4 and John Lintner5 independently developed a model so called Capital Asset Pricing Model as an extension to the Markowitzs optimization. Despite Linters argument that Professor Sharpe's paper, "Capital Asset Prices: A Theory of Market Equilibrium Under Conditions of Risk" (Journal of Finance. September 1964) appeared after his paper was in final form and on its way to the printer, Sharpe received the Nobel Price with Markowitz and Miller for his contribution to Capital Asset Pricing in 1990. Capital Asset Pricing Model, by Sharpe words, sheds considerable light on the relationship between the price of an asset and the various components of its overall risk. By having a common stock, there are two alternative to make profit which are capital gain and dividend. By capital gain, we mean the difference when you buy a stock and sell it. If you sell a stock on price higher than you bought, then difference is your capital gain. Second alternative is the amount of possible dividend you get from the company since you have a partial ownership in that company. It is called dividend. Some companies do not pay any dividend due to the fact that they are not obliged to do. However, some believe that to maintain a stable dividend pay-out ratio makes the stock valuable and protect company from the unfriendly takeovers. On the other hand, the amount of dividend is one of the reasons to buy a stock at private investor point of view. The aim of this paper to explore the risk & return relationships of banks common stocks in the context of Modern Portfolio Theory and Capital Asset Pricing Model. Literature Review The process of selecting a portfolio may be divided into two stages. The first stage starts with observation and experience and ends with beliefs about the future performance of available securities. The second stage starts with the relevant beliefs about future performances and ends with the choice of portfolio6 . Harry Markowitz, one of the pioneers of modern portfolio theory, said that the concept of yield and risk appear frequently in financial writings. Usually if the term yield were replaced by expected yield or expected return and risk by variance of return little change of apparent meaning would result. Modern Portfolio Theory Markowitz theorized that a properly diversified portfolio would provide maximum return for a given level of volatility or minimum volatility for a given level of return. Under the assumption of rational and risk averse investors, his approach centered on the belief that the desired output from an investors portfolio is a high return, while the cost to be minimized is the volatility (risk/standard

Holton,2004 3 Markowitz, H. 1952 4 Sharpe, W. 1964 5 Lintner, J. 1965 6 Markowitz, 1952

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE

562

deviation) of that return. Modern portfolio theory is therefore based on the trade-off between risk and return in a portfolio context7 . The risk and return are the main consideration to investment theory and its applications in real practices. The relationship between risk and return is positively correlated which means the higher expected return, the higher the risk or vice versa. In fact, risk is a measure of what is expected to happen, not what is actually happening. Expected return is the sum of each possible return that are calculated as each possible return times the probability of each state. This estimation is considered as an unknown future return which we define as expected return. In case of various state of economy, there might be huge difference among possible returns whereas expected return is the best estimation among others. Furthermore, what is needed to know in order to measure the risk is to calculate the spread of expected return around overall expected mean which is usually measured by the standard deviation or its square, variance. The expected return of a portfolio is just a weighted average of the each asset expected return taken place in that portfolio. The weights are the proportions of these assets held in the portfolio. When time comes to measure portfolio risk, we need to know one more statistical concept so called correlation coefficient denoted by p (in some text books it is labeled as r) which shows the joint movements between two variables, how they fluctuate. Correlation coefficient ranges between -1,0 to +1,0. if there is no correlation among two variables, we have a correlation coefficient close to 0. (at 0 there is no correlation at all). The magic tools of Modern Portfolio Theory are diversification which is totally a result of correlation of assets in portfolio. The risk is at its maximum when all correlations are perfectly correlated (+1). In this case, assets are substitutes for each other, which mean that as one of the variables increase or decrease in value, so does the other in exact proportions. The risk of the portfolio is the weighted average of the risks of the assets in the portfolio. When the correlation is -1.0, the return are perfectly negative correlated meaning that as one of the variables increase or decrease in value, so does the other in opposite proportions. The correlation coefficient for assets without any correlation at all is zero8 .
Figure 1: Correlation Analysis

Markowitz approach to portfolio diversification measured the proportion of funds to be invested in different individual assets and a measure of the expected return and the risk. The first stage was to compute the expected return and risk of each individual asset and to use these to calculate the portfolio expected return and risk from all possible combination of weights, using both linear programming and investing9 . In real world, it is not possible to find two perfectly negative or/and
7 8 9

Sandberg,2005 Perold, 2004 Hoesli, M., and MacGregor, B. D., 2000

563

aban elik

positive correlated assets since the factors that affect those assets vary time to time. Unless all of the correlations are perfectly correlated, the risk is always less than the weighted average. In such case, at least some of the risk from one asset to some extent will be offset by the other asset, so the standard deviation of the portfolio is always less than the average risk of the weighted average of the individual standard deviation. This consequence is lie behind diversification and portfolio construction. The amount by which risk is reduced depends on the correlations between the assets. The lower the correlation is between returns on different assets (the further away that the correlation is from +1), the greater the benefits of diversification10 . There is a possibility of reducing unsystematic risk as long as correlation coefficient among assets is lower than 1,0. The notion that diversification reduces risk is old. In the eighteenth-century English language translations of the roman Don Quixote, Sancho Panza advises his master, It is the part of a wise man tonot venture all his eggs in one basket11 . Markowitz succeeded to show analytically how the benefits of diversification depend on the extent to which individual asset returns are correlated. If Sancho Panza was to restate his advice according to Markowitz portfolio theory, he might instead say: It is safer to spread your eggs among imperfectly correlated baskets than to spread them among perfectly correlated basket12 . Imperfectly correlated assets act to insure one another. Capital Asset Pricing Model (Capm) Building on the Markowitz framework, Sharpe (1964), Lintner (1965) and Mossin (1966) independently developed what has come to be known as the Capital Asset Pricing Model. CAPM describes how investors determine expected returns, and thereby asset prices of risky assets, based upon their volatility relative to the market as whole. Just like Markowitzs Modern Portfolio Theory model, the CAPM is based upon several simplifying assumptions that make the model more tractable from a mathematical standpoint. For example, the CAPM assumes13 : Investors are rational and risk averse and tries to maximize their expected return over a single period. Investor only relies on two factors making their decisions: expected return and variance. Investors will ascribe to Markowitzs methodology of reducing portfolio risk by combining assets with counterbalancing correlations. Capital markets are completely competitive and frictionless. Investors are price takers and there are no taxes on transaction costs. There exists a risk-free asset in which there can be unlimited investment or borrowing at the same risk-free rate. Inflation is fully anticipated in the rate. All information is free and simultaneously available to all investors. The conclusions following from the assumptions are consequently14 : There is borrowing and lending at a risk-free rate, which is the same for all investors. Each investors portfolio of risky assets has the same composition as all other investors. The market portfolio, which is the aggregate of all portfolios, therefore has this same composition. The market portfolio is efficient for all investors, the unique mutual fund of all risky assets exactly suits the needs of all investors. Since the market portfolio is efficient, any other portfolio of risky assets is inferior. Sharpe, Lintner and Mossin demonstrated that the opportunity to borrow or lend at a risk freerate implies a capital market where risk-averse investors will prefer to hold portfolios consisting of
10 11 12 13 14

Perold, 2004 Perold, 2004 Sandberg,2005 Fabozzi, F. J., 1999 Sandberg,2005

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE

564

combinations of the risk-free asset and some portfolio on the efficient frontier15 . With a risk-free asset included, the Markowitz efficient frontier is no longer the best that investors can do. The straight line in the figure, the Capital Market Line, which has the riskfree rate as its intercept and is tangent to the efficient frontier, portfolio M, is now the highest boundary of the investment opportunity set.
Figure 2: Capital Market Line (CML)

Comparing a portfolio on the capital market line to a portfolio on the efficient frontier with the same risk, it is notable that for the same risk the expected return is greater everywhere along the capital market line (except for the tangency point, portfolio M, which is on the efficient frontier). Higher indifference curves can then be reached. With the introduction of the risk-free asset, an investor would instead of selecting a portfolio at the efficient frontier, select a portfolio on the capital market line representing a combination of borrowing and lending at the risk-free rate and a Markowitz efficient portfolio16 . But how does an investor find/construct this portfolio M? Eugene Fama answered this question by demonstrating that M must consist of all assets available to investors and each asset must be held in the proportion to its market value of all assets17 . If the total market value of some assets is X and the total market value of all assets is Y, then the percentage of the portfolio that should be allocated to that asset is X divided by Y. Because M consists of all assets, it is referred to as the market portfolio. The theoretical result that all investors will hold a combination of the risk-free asset and the market portfolio is known as the two-fund separation theorem18 - one fund consists of the risk-free asset, and the other is the market portfolio. The particular combination depends on the risk and return required by the investor. The relationship, can be expressed as a linear function of expected return on the market as a whole, in the following way19 :

Ri is the expected return on asset i over the period t. i is a term that represents the non-market component of the return on asset I (the risk free rate), and i is the term that relates the change in assets is return to the change in the return of the market portfolio Rm. Beta, measures the sensitivity of asset returns to the return on the market portfolio. The capital asset pricing model is a pricing model.

15 16 17 18 19

Fabozzi, F. J., 1999 Sandberg,2005 Fama, E. F., 1970 Tobin, J, 1958 Hoesli, M, and MacGregor, B. D., 2000

565

aban elik

If the beta of the investment and the expected market return are known, the expected return on the asset can be calculated and compared against the return which the asset is priced to deliver. A directly similar relationship also holds for individual assets expected return and its risk, which is known as the securities market line. Instead of expected return being a function of risk as measured by the standard deviation in the capital market line, it is now a function of the covariance (beta), , between the asset and the market. Assets lying above the security market line are under priced, since its expected return is above the level required by its beta. In contrast, asset lying below the security market line is over priced, since it delivers a return below that required by its beta20 . Research Methodology Based on Modern Portfolio Theory and Capital Asset Pricing Model, we analyzed nine Banks common stock from USA (Citigroup, bank of America corp., J.P. Morgan Chase & Co, Wells Fargo, Wachovia Corp, US Bankcorp, Capital One Financial, SunTrust Bank, National City Corp.) and nine from Turkey (Akbank, Denizbank, Finansbank, Garantibank, Is bank C, Sekerbank, T.S.K.B, Vakflar bankas, YKB) analyzing their daily return in 2006 independently. We consider S&P 500 in NYSE and ISE 30 as a market portfolio. Regression analysis was conducted for each stocks return with the market portfolios return in which market portfolio was employed as independent variable. Beta21 , as it is formulated in Sharpes paper, covariance of stocks return and market portfolios return divided by variance of market portfolios return. We made two different calculation and get the same results in order to measure systematic risk of each stock which is defined as the beta. Firstly, we used excel to make a regression for each stock with market portfolio and get the linear regression equation of which the beta is the coefficient of independent variable and R SQUARE which is the square of correlation coefficient22 , r for each stock and market portfolio return (table 1). Secondly, we computed (1)covariance of market return and stock return plus (2) variance of market return then divide 1 by 2 (table 3). Results are the same just like we computed through regression analysis. Since correlation coefficient is the key for diversification, we computed correlation matrix by employing SPSS in order to compare our result comes from excel application. It is important to note that we conducted both calculation based on daily stock returns whereas we also calculated monthly stock returns and find out that the betas coming from monthly returns analysis are lower than those of daily returns except two Turkish banks common stocks (T.S.K.B and YKB) which have higher betas coming from monthly returns than those of daily returns. On the other hand, we take split ratio factor into consideration before conducted the analysis23 . In order to see the effect of diversification, we assume that if we construct an equally weighted portfolios in two stock exchange, by how much we can diversify these two portfolios. As the results are demonstrated in table 6 and 10, the standard deviation of both portfolio are lower than all the stocks standard deviation.

20 21 22

23

Hoesli, M, and MacGregor, B. D., 2000 Sharpe, W. 1964 Bowerman et al. Forecasting, Time series and Regression.
split ratios: akbank(1,22:1); finansbank(1,31:1); is bank C(1,40:1); T.S.K.B(1,5:1); YKB (2,5:1); Wells Fargo (2:1)

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE
Table 1: Linear Regression equation for each stock:
Beta and R square y = 0,9243x + 0,0001 R2 = 0,4048 y = 0,8018x + 0,0002 R2 = 0,3912 y = 1,2717x + 0,0002 R2 = 0,5474 y = 0,7266x + 0,0003 R2 = 0,306 y = 0,9238x - 0,0002 R2 = 0,3034 y = 0,5523x + 0,0005 R2 = 0,2886 y = 1,1326x - 0,0009 R2 = 0,2297 y = 0,7726x + 0,0002 R2 = 0,3316 y = 0,9159x - 8E-05 R2 = 0,2938 Banks name akbank denizbank finansbank garanti bank is bank C sekerbank T.S.K.B vakflar bankas YKB

566

Banks name citigroup bank of ameraika corp. J.P. Morgan Chase & Co Wells Fargo Wachovia Corp. US Bankcorp Capital One Financial SunTrust Bank National City Corp.

Beta and R square y = 1,2707x + 0,0002 R2 = 0,7317 y = 0,4194x + 0,0022 R2 = 0,0854 y = 0,2206x + 0,0013 R2 = 0,0348 y = 1,2272x + 0,0001 R2 = 0,7456 y = 1,3525x - 0,0007 R2 = 0,7803 y = 0,7822x + 0,0004 R2 = 0,174 y = 1,2457x - 0,0003 R2 = 0,5412 y = 1,2723x - 3E-0 R2 = 0,6583 y = 0,9036x + 0,0002 R2 = 0,466

2. A Portfolio of Banks Common Stocks in ISE


Despite so many factors that affect a stock and market volatility, in the scope of this paper we deal with two main measurement techniques which are variance and standard deviation as an indicator of total risk and beta as an indicator of systematic risk. Therefore, no matter which stock we are taking into our portfolios, we are going to focus on its variance, standard deviation and beta. As we mentioned earlier, variance and standard deviation demonstrate the risk that relates that specific stock in the market. A beta, however, shows that stock covariability with the market portfolio return. In the following table, there are nine biggest banks which are entitled in ISE 100, 50, and 30 Indexes in Turkey in terms of their market capitalization. As it is seen that total share of those banks represent 41 % of ISE 100 Index, 45% of ISE 50 Index and almost 54 % of ISE 30 index. We also know that ISE 30 Index contributes nearly 76 % of ISE 100 Index24 . This is why in case of determining market portfolio, we are going to consider ISE 30 Index as a market portfolio just like S&P50025 .
Table 2: key statistics for common stocks in portfolio from ISE
capital in paid (TRY) 2.400.000.000 316.100.000 1.250.000.000 2.100.000.000 2.756.585.000 125.000.000 300.000.000 2.500.000.000 3.142.818.454 14.890.503.454,00 EPS 14,7 19,5 9,4 13,4 15,5 30,6 8,29 11,3 NA Current Price (*) 9,20 13,70 6,10 6,70 6,25 8,40 2,10 3,84 3,00 52-Week Change %10,51 %-1,44 %3,39 %44,16 %-1,66 %76,47 %8,26 %12,85 %22,36

common ISE 100 ISE 50 ISE 30 % shares market stock outstandings capitalization ($) 8,98 9,95 11,79 27,33% 16.664.150.943 akbank 1,54 1,70 2,02 25,00% 3.268.354.717 denizbank 4,89 5,41 6,41 46,76% 5.707.547.170 finansbank 8,27 9,15 10,84 51,61% 10.539.622.642 garanti bank 9,05 10,03 11,88 33,18% 13.002.759.434 is bank C 0,92 1,02 1,20 97,49% 792.452.830 sekerbank 0,72 0,80 0,95 54,03% 633.962.264 T.S.K.B 3,31 3,66 4,34 12,88% 7.245.283.019 vakflar bankas 3,33 3,68 4,36 25,63% 6.847.988.262 YKB TOTAL 41,00 45,41 53,79 62.730.431.083,00 Source: http://www.imkb.gov.tr/bultenler.htm and http://piyasanet.hurriyet.com.tr (*) date: 21 may 2007

Table 2 demonstrates some key statistics for common stocks which we want to learn their systematic and unsystematic risk indicators for the year of 2006.

24 25

http://www.imkb.gov.tr/bultenler.htm there are two reasons lie behind the decision why we choose ISE 30 Index as a market portfolio. Firstly, it represents whole market as much enough as we need. (because S&P500 represents more or less by the same amount in NYSE). Secondly, we wanted to see what is going on among top companies comparing with the other biggest companies in Turkey by focusing on Banking Industry.

567 Interpretation of key formulas and indicators

aban elik

In the following table (table 3), as it is seen that the betas as an indicator of systematic risk and variances and standard deviations as an indicator of total risk26 presented for each stock. We collected daily data for each stock from Istanbul Stock Exchange Website and make our own calculation in order to gain the following inferences. We compare the returns of market portfolio return which is ISE 30 Index in this case and return for each stock. In order to calculate beta, since beta27 is covariance of market return and a stock return divided by market variance, we calculated covariance of market return and each stock return and then divided covariance of market return and a stock return by variance of market return. As a result we get the beta for each stock. As it is demonstrated in table 3 and 8, correlation coefficient is represented as r 28and R square is just a square of correlation coefficient r. As we mentioned earlier, correlation coefficient shows how a stock return moves in the direction of market portfolio return. Additionally we were wondering how those stocks are correlated between each other. Thats why we make a correlation matrix which represented in table 5 and 9 . On the other hand, R square emphasizes how much our data are represented in the regression.
Table 3 :
ISE 30

The Result of Analysis for Common Stock in Portfolio from ISE


akbank denizbank finansbank 0,00008 0,0004 -0,4403 0,2197 0,1865 0,0348 0,0005 0,0228 0,2989916 garanti bank 0,0005 0,0004 1,1597 1,2223 0,8635 0,7456 0,0008 0,0274 -0,05645 is bank C 0,0005 0,0004 1,253 1,347 0,883 0,78 9E-04 0,03 -0,23529 sekerbank 0,0003 0,0004 -0,113 0,779 0,4172 0,174 0,0013 0,0362 -0,0574 T.S.K.B 0,0005 0,0004 1,7945 1,241 0,736 0,541 0,001 0,033 -0,1766 vakflar bankas 0,0005 0,0004 1,05409 1,2672 0,8114 0,6583 0,0009 0,0302 -0,1073826 YKB 0,0003 0,0004 1,1844 0,9 0,6826 0,466 0,0007 0,0255 -0,0238

0,0005 0,0002 cov(k,km) 0,0004 0,0004 var(km) 0,97616 -0,6598 Beta(*) 1,2656 0,4177 beta 0,8554 0,2922 r 0,7317 0,0854 R square 0,0008 0,0008 var(ki) 0,0286 0,0277 std D(ki) -0,0374 0,58857 rate of return -0,039 rate of return ISE 30 (*) the betas computed on monthly returns.

Inferences of the analysis Hopefully, we succeeded to come the most exciting point where we interpret what we get from our analysis. We have been analyzing common stocks of companies that are actively operating in Turkish banking sector. In the scope of our analysis, we calculated systematic and unsystematic risk indicators for these stocks in the period of one year, 2006. Let us start to evaluate our results step by step: Akbank is the biggest bank in Turkey in terms of market capitalization. When we look at its common stock performance in 2006, we find out that its return covariability with market portfolio (ISE 30 Index) is 1,3 which we called Akbanks beta in 2006 based on daily return data. The conclusion comes out from this number is that Akbanks common stock carries 1,3 as volatile as the market and therefore it carries 1,3 as much risk. At Akbanks total risk point of view, we are looking at its return variance and standard deviation as an total risk indicators. As it is demonstrated that Akbanks variance and standard deviation are 0,00082 and 0,03 respectively. 2,9% standard deviation is almost an average of nine common stocks average. There is remarkable conclusion we drawn from monthly return of banks common stocks that betas from monthly returns are lower than daily returns29 . Standard deviation 30 should be evaluated with others stocks standard deviations in order to make a judgment.
26 27 28 29 30

Total risk= systematic risk (market risk) + unsystematic risk (company specific risk or diversifiable risk) Sharpe, W. 1964 this r usually called pearson correaltion coefficient Kothari, Shanken, and Sloan (1995) claim that s from annual returns produce a stronger positive relation between and average return than s from monthly returns (Fama, French,1996) variance is expressed in terms of squared units, it is difficult to interpret, that is why analysts often convert variance to an expression of standard deviation as a more convientient measure of total risk.

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE

568

We may say that Akbank carries on average total risk among others. (total risk31 is expressed as the sum of systematic and unsystematic risk)
Table 4: Total Risk versus Systematic Risk
Standard deviation Total risk 0,02864 0,02767 0,0228125 0,02741 0,02952 0,036152 0,032652 0,0302371 0,02552 Beta(*) Systematic risk 1,3 0,4 0,2 1,2 1,4 0,8 1,2 1,3 0,9

Common stock akbank denizbank finansbank garanti bank is bank C sekerbank T.S.K.B vakflar bankas YKB
(*) Betas are rounded to the nearest 0,50

From our discussion about beta and total risk, we may say that Akbank has a higher total risk than Vakflar Bank, even though their systematic risk is equal. Akbank, Isbank, Garanti Bank, T.S.K.B. and Vakflar Bank have a greater systematic risk relative to that of an average risky asset. We draw a remarkable conclusion that the reward for bearing risk depends only on the systematic risk of an investment according to systematic risk principle32 . Therefore, Isbank Bank has a higher risk premium and greater expected return than others. At private investor point of view, beta is usually signal of stock which shows to degree of rewarding to investors risk tolerance. In this case we may say that Denizbank, Finansbank, Sekerbank and Yap Kredi Bank have lower systematic risk among others. However, beta does not capture the unsystematic risk. That is why we also take standard deviation of a stock into account before deciding to pick up that stock into our portfolio. In general, common stocks we analyzed made a very poor performance in 2006. Except Denizbank and Finansbank, all banks made a negative rate of return just like market portfolio. However, we found out that systematic risk indicators, beta33 , does not explain what it supposed to be. In case of 58% positive return generated by Denizbank, we see that denizbanks beta is 0,4 which is less than average whereas it reached the highest return among those we analyzed. It is the same for almost all common stocks at some degree. The conclusion comes out from this results is either those stocks are not fairly valued or market is not well-functioning, competitive market. (Remember what are the assumptions lie behind the CAPM34) At diversifying portfolio point of view, we have to look at correlation coefficient between market portfolio return and stocks returns. The lowest correlations performed by Denizbank and Finansbank, not surprisingly those which made positive returns among others. The critical question is whether or not is there any negatively correlated stocks return in same sector? In case of analyzing banks common stocks in this paper, as it demonstrated in correlation matrix at the following table, we could have found that Finansbank is negatively correlated with T.S.K.B and YKB stocks. This is not

31 32

33

34

The total risk of an investment is measrued by the standard deviation of its return. (Ross et al: Fundamentals of corporate, page 458) the underlying rationale for this principle is straightforward: because unsystematic risk can be eleminated at virtually no cost (by diversifying), there is no reward for bearing it. Thats why the expected return on an asset depends only on that assets systematic risk. (Ross et al: Fundamentals of corporate, page 458) the beta coefficient measures the covaribility (relative movement) of a stock as compared to a benchmark which, in this case, market portfolio. An average risk stock has a beta of 1. this means that if the market moves up by 10%, the stock will, on average, move up by 10% or vice versa. Less than average risk stocks have betas between 0 and 1 whereas higher than average risk stocks have greater than 1. (Citigroup) Fama and French conclude that the commonly accepted relationship between beta and return does not exist at all. people have found that the correlation between higher beta and higher returns is much flatter than expected by CAPM. We found it to be completely flat. (Sandberg,2005)

569

aban elik

supposed to happen in the same sector. The lower correlation coefficient means the higher diversifiable portfolio.

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE
Table 5: correlation matrix for common stocks in portfolio from ISE
ISE30 AKBANK FINAN SBANK 0,113 0,046 ,323(**) 1 0,086 0,050 0,085 -0,035 0,014 -0,058 GARAN TIBANK ,863(**) ,662(**) ,231(**) 0,086 1 ,682(**) ,344(**) ,567(**) ,694(**) ,381(**) ISBANK C ,754(**) ,597(**) ,160(*) 0,050 ,682(**) 1 ,287(**) ,479(**) ,608(**) ,306(**) SEKER BANK ,417(**) ,284(**) ,236(**) 0,085 ,344(**) ,287(**) 1 ,296(**) ,283(**) ,163(*) TSKB ,693(**) ,554(**) ,183(**) -0,035 ,567(**) ,479(**) ,296(**) 1 ,578(**) ,315(**) VAKIFL ARBANK ,811(**) ,669(**) ,173(**) 0,014 ,694(**) ,608(**) ,283(**) ,578(**) 1 ,323(**)

570

DENIZ BANK ISE30 r 1 ,813(**) ,292(**) AKBANK r ,813(**) 1 ,212(**) DENIZBANK r ,292(**) ,212(**) 1 FINANSBANK r 0,113 0,046 ,323(**) GARANTIBANK r ,863(**) ,662(**) ,231(**) ISBANKC r ,754(**) ,597(**) ,160(*) SEKERBANK r ,417(**) ,284(**) ,236(**) TSKB r ,693(**) ,554(**) ,183(**) VAKIFLARBANK r ,811(**) ,669(**) ,173(**) YKB r ,432(**) ,281(**) 0,073 **. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed). Own calculation by using SPSS

YKB ,432(**) ,281(**) 0,073 -0,058 ,381(**) ,306(**) ,163(*) ,315(**) ,323(**) 1

We come to the point where to show how diversification works. In table 6, there is a correlation matrix available and the following table demonstrates the rate of return of stocks taken place in an equally weighted portfolio.
Table 6:
common stock % of total day 1 day 2 day 3 average var std

the effect of diversification on the portfolio from ISE


akbank 0,11 0,0051 0,001 0,0000 0,0000 0,0000 0,0032 denizbank 0,11 0,0051 0,0012 0,009 0,0003 0,0000 0,0031 finansbank 0,11 0,006 0,007 0,008 0,0002 0,0000 0,0025 garanti bank 0,11 0,004 0,002 0,003 0,0000 0,0000 0,0031 is bank C 0,11 0,0037 0,0036 -0,003 -0,0001 0,0000 0,0033 sekerbank 0,11 0,006 0,001 0,004 0,0001 0,0000 0,0040 T.S.K.B 0,11 0,0033 0,0005 0 0,0000 0,0000 0,0036 vakflar bankas 0,11 0,00373 0,00289 -0,0021 0,0000 0,0000 0,0034 YKB 0,11 9E-04 9E-04 9E-04 0,0000 0,0000 0,0028 total 1 0,037 0,02 0,021 0,0004 0,0004 0,0195

As it is realized that the standard deviation of an equally weighted portfolio is 0,0195 which is much lower than average of standard deviation of common stocks return presented in table 3. We have to underline the fact that we construct the portfolio equally weighted as it is demonstrated in table 6 (11% for each). This is the effect of diversification. The portion that we can not diversify is our portfolios systematic risk that is not diversified in this circumstance. We have to add up more common stocks if we want to decrease our portfolios unsystematic risk especially from those which are negatively correlated.

3. A Portfolio of banks common stocks in NYSE


In case of interpreting what happened in portfolio which contains nine biggest banks from Down Jones, we are going to evaluate results without explaining its theoretical relations since we have already explained it in detail in previous section. That is why we will satisfy to show the results of our analysis:

571
Table 7:
Ticker C BAC JPM WFC WB USB COF STI NCC source:

aban elik
key statistics for common stocks in portfolio from NYSE
list of banks in USA citigroup bank of ameraika corp. J.P. Morgan Chase & Co Wells Fargo Wachovia Corp. US Bankcorp Capital One Financial SunTrust Bank National City Corp. http://finance.yahoo.com (*)date: 21 may 2007 (**) yahoo calculation Market Cap 272.05B 227.57B 179.48B 121.63B 108.41B 60.47B 32.18B 31.92B 20.57B P/E 13.13 10,94 11,62 14.28 11,96 13.36 12,06 15.46 10,30 Shares Outstanding 4.95B 4.44B 3.42B 3.34B 1.91B 1.74B 415.59M 356.62M 575.76M 52-Week Change 12.70% 6.24% 22.90% 10.26% 6.62% 11.83% -7.51% 18.34% -2.11% S&P500 52Week Change 20.18% 20.18% 20.18% 20.18% 20.18% 20.18% 20.18% 20.18% 20.18% Beta(**) 0,59 0,32 0,69 0,35 0,53 0,48 1,01 0,2 0,55 Current Price(*) 55.08 51.50 52.29 36.00 56.25 34.42 78.62 90.61 35.57

Table 7 demonstrates some key statistics for common stocks we have been working on. The remarkable change from the portfolio we analyzed in previous section is the huge difference occurred in market capitalization stand point. As it is seen that betas are already calculated for those stocks of which we suffered to calculate it for those common stocks from ISE. This is one of the indicators of well-functioning financial markets. Unfortunately, in Turkish financial markets, there is huge asymmetric information which creates insufficient market. Since we have ready calculated beta from finance.yahoo.com, we are going to compare what would be the difference in terms of time and the difference we expect to occur. Those banks are listed in fortune 500 in the rank of biggest bank in USA. For this respect we are going to see what happened in banking sector among the biggest player in the world.
Table 8:
S&P500

The result of analysis for common stock in portfolio from NYSE


citibank bank of ameraika corp. J.P. Morgan Chase & Co 0,00005 0,00004 0,82288 1,26663 0,73985 0,54738 0,00012 0,01074 0,20 Wells Fargo 0,00003 0,00004 0,07383 0,72374 0,55318 0,30601 0,00007 0,00821 0,17 Wachovia Corp. 0,00004 0,00004 0,56800 0,92009 0,55085 0,30343 0,00011 0,01048 0,05 US Bankcorp 0,00002 0,00004 0,38663 0,55008 0,53719 0,28857 0,00004 0,00642 0,20 Capital One Financial 0,00004 0,00004 0,13014 1,12806 0,47925 0,22968 0,00022 0,01476 -0,12 SunTrust Bank 0,00003 0,00004 0,24340 0,76956 0,57588 0,33164 0,00007 0,00838 0,14 National City Corp. 0,00004 0,00004 0,05828 0,91219 0,54200 0,29377 0,00011 0,01056 0,07

cov(k,km) 0,0000 0,00003 var(km) 0,0000 0,00004 Beta(*) 0,5634 0,42816 beta 0,9206 0,79859 r 0,6362 0,62545 R square 0,4048 0,39119 var(ki) 0,0001 0,00006 std D(ki) 0,0091 0,00801 rate of return 0,13 0,13 rate of return S&P 500 0,11783 (*) the betas computed on monthly returns.

Table 8 shows the fundamental results of our analysis which we made for portfolio that contains nine biggest banks common stocks. As it is realized that all betas we calculated are higher than those which are calculated by yahoo. The reason is simple that we limited the time period for the data we worked on it. Second reason is that we used daily return for stocks which makes our analysis more sensitive. When we look at the result, we see that the highest systematic and total risk carried by Capital One Financial. Not surprisingly it has the highest beta in the list of yahoo. The remarkable conclusion as we gained in the previous section is that these betas do not support the CAPMs relationship with the return assumption. As you see that Wells Fargo with a beta of 0,7 contributed higher return that market portfolio, just like US Bancorp did. One of the important inferences is also the movements of those stocks with the market return. As it is realized that correlation coefficients of those stocks are very close to each other which underline the fact that they response almost the same reaction to the factors influencing banking sector in USA.

Risk & Return Relationship in the Portfolio of Banks Common Stocks Nyse Versus ISE
Table 9: Correlation matrix for common stocks in portfolio from NYSE
SP500 Jp morgan ,740(**) ,663(**) ,696(**) 1 ,680(**) ,629(**) ,536(**) ,415(**) ,515(**) ,601(**) Wells fargo ,553(**) ,573(**) ,665(**) ,680(**) 1 ,662(**) ,583(**) ,352(**) ,491(**) ,596(**) Wachovia Corp ,551(**) ,575(**) ,568(**) ,629(**) ,662(**) 1 ,541(**) ,298(**) ,480(**) ,532(**) Us bank corp ,537(**) ,481(**) ,534(**) ,536(**) ,583(**) ,541(**) 1 ,358(**) ,566(**) ,566(**) Capital one fin ,479(**) ,330(**) ,304(**) ,415(**) ,352(**) ,298(**) ,358(**) 1 ,234(**) ,355(**) Sun trust bank ,576(**) ,519(**) ,524(**) ,515(**) ,491(**) ,480(**) ,566(**) ,234(**) 1 ,566(**)

572

Citi Bank of group amrica SP500 r 1 ,636(**) ,625(**) Citigroup r ,636(**) 1 ,614(**) Bankofamrica r ,625(**) ,614(**) 1 Jpmorgan r ,740(**) ,663(**) ,696(**) Wellsfargo r ,553(**) ,573(**) ,665(**) Wachovia r ,551(**) ,575(**) ,568(**) Usbancorp r ,537(**) ,481(**) ,534(**) Capitalonefin r ,479(**) ,330(**) ,304(**) Suntrustbank r ,576(**) ,519(**) ,524(**) Nationalcity r ,542(**) ,501(**) ,548(**) **. Correlation is significant at the 0.01 level (2-tailed).

National city ,542(**) ,501(**) ,548(**) ,601(**) ,596(**) ,532(**) ,566(**) ,355(**) ,566(**) 1

Among common stocks we analyzed, as it is seen that there is not any negatively correlated stocks in banking sector of USA. This is what we expected due to the fact that returns of those stocks move in the same direction at some degree whereas the range is between 0,3 and 0,6. This figure draws an inevitably significant conclusion that we will have a certain limit in diversifying this portfolio. Even though those correlation coefficients are quite close to each other, we still have a place to diversify our portfolio. It is not a big surprise to see that correlation coefficient of stocks with the market return is quite close to each other. This shows that the tendencies of those stocks are in the same direction at some degree of difference. The lowest correlation observed on Capital One Financials stock while the highest one is J.P. Moregans correlation coefficient. In other words, J.P. Morgan move in the tendency of market return by 0,7. We have to remind you that correlation coefficient does not explain causality effect between returns. It just shows the direction of their movements. We are going to show how this diversification works in table 10.
Table 10: The effect of diversification on the portfolio from NYSE
Common stocks % of total day 1 day 2 day 3 average var std Citibank Bank of ameraika corp. 0,11 -0,001 0,000 0,000 0,000 0,000 0,001 J.P. Morgan Chase & Co 0,11 -0,002 0,000 0,001 0,000 0,000 0,001 Wells Fargo 0,11 -0,001 0,000 0,001 0,000 0,000 0,001 Wachovia Corp. 0,11 0,000 -0,001 0,001 0,000 0,000 0,001 US Bankcorp 0,11 0,000 0,001 0,001 0,000 0,000 0,001 Capital One Financial 0,11 -0,001 0,000 0,000 0,000 0,000 0,002 SunTrust Bank 0,11 0,001 0,000 0,000 0,000 0,000 0,001 National City Corp. 0,11 0,001 0,001 -0,001 0,000 0,000 0,001 Total

0,11 -0,002 0,001 0,000 0,000 0,000 0,001

1 -0,004 0,002 0,003 0,000 0,000 0,007

As we did in previous section, we construct an equally weighted portfolio investing in each stock by 11%. In order to see how diversification works, we have to compare the standard deviation of stock returns with the standard deviation of portfolio return. As it is seen that the standard deviation of our portfolio is 0,007 which is lower than the lowest standard deviation we calculated for common stocks in portfolio. This is an important outcome as we mentioned earlier for the purpose of reducing unsystematic risk of those stocks. We should not forget that this is not the limit of diversification we can make. In fact, we have a limit of choosing stocks and analyze their contribution for this portfolio due to the limitation of this research. Actually, we would see, if we picked up some stocks (especially those which have negative correlation with the stocks already placed in portfolio) into our portfolio. In general, a well-diversified portfolio can be constructed by 40 stocks on average.

Conclusion
We analyzed banking sector in USA and TURKEY focusing on 18 biggest banks common stocks and measured their risk return relationships in context of Modern Portfolio Theory and Capital Asset Pricing Model. The study indicates that risk return relationships of these stocks are not fully explained

573

aban elik

by CAPM. Additionally, the stocks movements of common stocks from USA react in the same direction (there is no negative correlation coefficient among those stocks returns and their correlation coefficient with the market returns are quite close to each other) whereas we could not have observed the same movements among common stocks from TURKEY (there is one common stock which is negatively correlated with two other stocks and the spread of their correlation coefficient with the market return are larger than those from USA.). Market risk indicator, beta, of Turkish Banks common stocks is much higher than those of USA. Volatility of American Banks common stocks are lower than those of Turkish. Consequently, all the signals we have observed by comparing this two portfolio shows that Capital Market in USA is more efficient than capital market in Turkey. However, we had better opportunity to make more diversified portfolio in case of Turkish Banks common stocks due to the fact that the spread of their correlation coefficient among themselves.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] Fundamentals of Corporate Finance, Stephen A. Ross, Randolph W. Westerfield, Bradford D. Jordan, Sixth Edition, Published by McGraw-Hill/Irwin,. Basics of Corporate Finance, / citibank, Latin America Training and Development Center, May 1994, self-instruction series. Forecasting, Time Series, and Regresion, fourth edition, / Bowerman, OConnell, Koehler Hoesli, Martin. and Bryan D. MacGregor, 2000, Property Investments Principles and Practice of Portfolio Management, Longman - an imprint of Pearson Education. Fabozzi, Frank J., 1999, Investment Management, Pearson Education/Prentice Hall, New Jersey, Second Edition. Sharpe, William F. (1964). Capital asset prices: A theory of market equilibrium under conditions of risk, Journal of Finance, 19 (3), 425-442. Markowitz, Harry M. (1952). Portfolio selection, Journal of Finance, 7 (1), 77-91 Fama, Eugene F., 1970, Efficient Capital Markets: A Review of Theory and Empirical Work, Journal of Finance, May 1970, pages 383-417. Fama, Eugene F.,1996. The CAPM is Wanted, Dead or Alive. The Journal of Finance, Vol. 51, No. 5. (Dec., 1996), pp. 1947-1958. Hal Varian, A Portfolio of Nobel Laureates: Markowitz, Miller and Sharpe, The Journal of Economic Perspectives, Vol. 7, No. 1. (Winter, 1993), pp. 159-169. Sandberg, Peter(2005). How Modern is Modern Portfolio Theory, master thesis, Royal Institute of Technology, Stockholm 2005. Lintner, John (1965). The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets. The Review of Economics and Statistics, Vol. 47, No. 1. (Feb., 1965), pp. 13-37. Perold, Andr F., 2004, The capital asset pricing model, Journal of Economic Perspectives, vol. 18 number 3, pages 3-24. Tobin, James, 1958, Liquidity Preference as Behavior towards Risk, Review of Economic Studies, February 1958, pages 65-86. Istanbul Stock Exchange official web site: http://www.imkb.gov.tr/bultenler.htm Yahoo. Finance: http://finance.yahoo.com/

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.574-584 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Ethnic Violence in Sindh and its Repercussions on South Asia


Amir Ali Chandio Assistant Professor Political Science Shah Abdul Latif University, Khairpur Pakistan Ishtiaq Ahmed Choudhry Dean Faculty of Arts, Social Sciences & Law University of Sargodha Sargodha Pakistan. Kilometers Sindh Quarterly, Vol. XXII (1994) No.4, p, 59 Karachi Abstract Ethnic violence in Pakistans Southern Province Sindh left many repercussions on South Asian politics and the relations of the neighboring states. In this article emphasis would find the causes of the ethnic violence and its impact on the politics of the country as well as to search its repercussions on the south Asian politics and relations. Before the independence there was no any ethnic conflict. There were no quarrels and tensions between the residences. The residence lived with love and peace. The contradiction grew among the residence of the province after creation of Pakistan due to the dominancy of the migrated people. The ethnic violence became in serious under the rule of military junta. First serious ethnic violation happened between the Pashtoons and Muhajirs under the rule of Ayoub khan in 1960s. After it the ethnic riots took place between the Sindhis and Muhajirs in 1972. The very serious ethnic riots started during Zia government in 1980s. First riots erupted between the Muhajirs and Pashtoons and than Punjabis and Muhajirs and later between the Sindhis and Muhajirs. Hundreds were killed in the ethnic riots. Riots left the various repercussions on the country as well as on regional politics. Due to those riots a tense situation was continuous among the groups. And also it is threat to the regional peace and security in South Asia.

Introduction
Sindh is the most important province of Pakistan due to its geopolitical situation. Sindh province is located in the southeast with the border of India and Arabian Sea. It is covered about 1,51,440 square kilometers area.1The total population of Sindh province was 30.440 millions in 1998.2 Sindh lies between 23-25 and 28-30 north longitude and 66-42 and 71-10 east longitude. It is about 579 kilometers in length from north to south and its extreme wideness is 442 kilometers. The average breadth is about 281 kilometers.3

1 2 3

Encyclopedia Americana, 24th edition 1986, p, 837 Pakistan Economic Survey 2004-05 P, 93, Government of Pakistan Islamabad June o4, 2005 Sindh Quarterly, Vol. XXII (1994) No.4, p, 59 Karachi

Ethnic Violence in Sindh and its Repercussions on South Asia

575

There are three main ethnic political powers in Sindh Province. First Sindhi and Baloch, they are in majority in province, but in minority in the capital city of the province. The second ethnic power in Sindh is the Muhajirs who were migrated from India and settled down in the urban areas such as Karachi, Hyderabad and Sukkur. They held important position in bureaucracy and occupied key posts in the government structure. Government of Pakistan declared Urdu as a National Language. Even it was the mother tongue of only 3% people of the country. The Sindhi language oppressed in Karachi, especially after the declaration as Federal Capital of the country. After the separation of Bengal from Pakistan, the Urdu speaking were 7% of the total population of the country and one fifth of Sindhs population.4 The third factor in the ethno political power in Sindh is the Punjabis and Pashtoons. After the formation of One Unit a huge number of Punjabis and Pashtoons settled in Sindh. The vast area of barrage land of Sindh was allotted to them and Jobs was also offered. The licenses of the business also issued to the Punjabis and Pashtoons. This phenomenon changed the geopolitical situation of Sindh. The status of Sindhi language was more reduced in the time of One Unit. People of Sindh recorded their protest against this discrimination, but military ruler did not respond them. The total population of the various ethnic groups in Sindh were as under in percentage according to 1981 Census.
Language Sindhi Urdu Punjabi Balochi Pushto Siraki Birahvi Hindko Others % Population 52.40 22.64 7.69 4.51 3.06 2.02 1.08 0.36 5.97

Sources: Census 1981 (Quoted by http://yangtze.cs.uiuc.edu/jamali/sindh/res/documents/mohpop.html

Urdu speaking people settled in Sindh even before the partition of India. They lived peacefully in Sindh and there was no tension among the residences of the province. First ethnic violence in the history of Pakistan erupted between the Muhajirs and Pashtoons under the authoritarian government of General Mohammad Ayoub Khan, in 1960s. The second ethnic riots were started in the 1972, between Sindhis and Muhajirs on language issue. A big series of the ethnic riots were occurred in Sindh under the military regime of General Zia, many innocent people were killed. The law and order situation of the province remained disturbed level. Therefore, capitalist and business class remained confused to open and establish new industries and business activities in the province. The political stability in the province remained at low ebb. This situation continued even in the governments of Benazir Bhutto and Nawaz Sharif. So the riots left many political impacts on country as well as on the politics of South Asia.

Punjabi Pashtoons and Muhajirs Riots


The first ethnic riots in Sindh province after the independence were started between the Pashtoons and Muhajirs during the authoritarian rule of Field Marshal Mohammad Ayoub Khan in the Presidential elections. Fatima Jinnah sister of Quaid-e-Azam Mohammad Ali Jinnah contested the presidential elections with Mohammad Ayoub Khan. Majority of the Urdu-Speaking people were supported Fatima Jinnah and Pashtoons were supporter of Ayoub Khan. The electoral collage of the presidential election was consisted on the elected members of the local bodies. Ayoub khan won the elections. After the
4

Froze Ahmed (2000) Ethnicity, state and national integration II Dawn April 12- 2000

576

Amir Ali Chandio and Ishtiaq Ahmed Choudhry

result of the elections a rally was organized by the Pashtoons under the leadership of the Son of President Ayoub Khan. When rally reached in Muhajir dominated areas the riots took place between the Pashtoons and Muhajirs. Once again ethnic riots took place between the Punjabi, Pashtoons and Muhajirs during military regime of General Zia, in 1985. The riots were continued more than two years. The Causes of the Riots The ethnic riots between the Punjabis, Pashtoons and Muhajirs were the result of the economic contradictions. In these riots Punjabis and Pashtoons were on one side and Muhajirs on other side. The riots were started after the green revolution of General Ayoub Khan, when a big number of Punjabis and Pashtoons selected the way of Karachi. The migrated Punjabis and Pashtoons settled at Karachi the capital city of Sindh province. The Muhajirs were dominated in the city since independence. The migration of the Pashtoons and Punjabis and other peoples from the various parts of the country, occupied the jobs in the city. So the chances of the employment for the Muhajirs were reduced. Pashtoons worked in the low level jobs such as unskilled labour in the industries, hotel Waiter and in transports. And others were involved in the drug mafia profession. Babar Ali narrated it in these lines The fact that country was governed by Pashtoons, Ayoub Khan also added to the security of Pashtoons and helped them migrate, as they sincerely believed that it was their government. Today the Pashtoons constitute the large section of the working class; especially in construction (i.e. unskilled laborers) they work in factories, and majority of transport workers also belong to them. Most of the Pashtoons are thus part of the lower class in Karachi. Some, however, made a great deal of money in the drug and arms trade in the city.5 The Afghan Refugees also came in Karachi during Afghan crises in Zia regime. Many of them were settled in the city, some of them supported the local Pashtoons due to same language. Punjabis were educated so they competed with the Muhajirs in the civil services as well as in the business and industries. Therefore, the contradiction took place between them was natural. Many Punjabis moved to Karachi in the search of work and as an industrial city there were more jobs as compare to the other cities of the country. Babar Ali wrote as under The large number of the Punjabis in Karachi attracted to the city because it was the capital (before 60s) and thus offered all the advantages accordingly. With the bureaucracy playing a major role in the earlier years in setting up business and industry, the Punjabi entrepreneurial community found it easier to settle in Karachi as it offered direct access to the bureaucracy. Thus a very great number of Chinioti businessmen moved to Karachi and are today one of the leading groups in industry in the country Apart from this upper class, smaller businessmen, from the Punjab also moved to the capital. (Karachi).6 Ayoub regime supported the Punjabis and Pashtoons as compare to the Muhajirs. The tension continued between them till the downfall of Ayoub government. The conflict reduced between the Punjabis, Pashtoons and Muhajirs during Z. A. Bhutto, government, because, during the Bhutto rule, the contradiction and dispute became high between the Sindhis and Muhajirs on the language issue. Punjabis and Pashtoons supported Muhajirs, and made the Punjabi Pashtoon Muhajir Mutahida Mahaz (PPMMM). Z. A. Bhutto was the first Sindhi, who succeeded to acquire the highest post of the country. He was trying to compensate the Sindhis in government as well as in the private sector. He accommodated many Sindhis. The Sindhi Language Bill passed by the Sindh Assembly and Quota System was extended. These were the main reasons of dispute between the Muhajirs and Sindhis. Muhajirs moved against Bhutto in 1977. The Muhajir dominated areas in Sindh were the centers of anti Bhutto movement. As a result of this movement, military intervened Martial Law was imposed. General Mohammad Zia-ul- Haq became the Chief Martial Law Administrator. It was the hopes of the

5 6

Babar Ali (1992) Political forces in Sindh Regional Imbalances and the National Question in Pakistan, P, 182, Vanguard Books Lahore Ibid, p181

Ethnic Violence in Sindh and its Repercussions on South Asia

577

Muhajirs that Military government would favour them. But their wishes went into wave when Martial Law regime favoured Punjabis as compare to Muhajirs Zia regime patronized the Punjabis in Sindh; the settler Punjabis of rural Sindh also migrated to Karachi. They availed important posts of the government especially in police and other law enforces agencies. Many Afghan Pashtoons also came in Karachi due to Zia's Afghan policy. The migration and settlement in Sindh especially in Karachi was the main reason of the contradiction and dispute between the Muhajirs, Punjabis and Pashtoons. The migration to wards Sindh Province also cleared from the total population of the Punjab and Sindh in Pakistan which as under
Year Punjab Sindh 1972 37,612 (57.6) 14,158 (21.7) 1981 47,292 (56.1) 19,029 (22.6) 1998 73,621 (55.62) 30,440 (23.00)

Population in Thousand: Sources: Pakistan Economic Survey, 2004-05, Government of Pakistan Islamabad June 4. 2005, P, 97].

The percentage of the population of the Punjab and Sindh Province showed that the percentage of Punjab reduced and the percentage of Sindh was increased. It did due the migration to Sindh. The death of Bushra Zaidi a Muhajir college girl died in Mini bus accident was the current reasons of the riots. The angry mob torched the mini bus and stoned the buses. Pashtoons were dominated on the transport of the city so; the riots erupted between the Muhajirs and Pashtoons. General K. M. Arif also agreed that the ethnic clash in Karachi between Muhajir and Pashtoons was the result of accident of Muhajir girl, he wrote that In April 1985, a Female student, Bushra Zaidi, was run over and killed by a mini- bus in a road accident. This unfortunate incident took an ugly ethnic turn because the bus driver was Pashtoon , and victim a Muhajir.7 MQM played the leading role in ethnic riots. This is a ethno nationalist organization of the Muhajirs which which were migrated from India and settled in Sindh. The Ethno Nationalism was the base of MQMs politics. When MQM succeeded to gain the support from urban areas of Sindh province, the ethnic riots continued one or other shape. The people of the Sindh divided on basis of language. The politics of ethnicity was grown up on those days and ideological politics became weak. Where from the rightist federal parties like JI and JUP lost their support. MQM started their practical politics from the fist public meeting in Nishtar Park at Karachi on 8th August 1986. It was marked by heavy aerial firing from pistols and riffles, which the party activists were carrying on them.8 The some of statements of Altaf Hussain also instigated the youth to take the arms. It was his address to press club Hyderabad on 25 October 1986, he told the Muhajir youth Collect arms, if our rights are not given to us, we will use every kind of force.9 The second big round of the riots erupted between the Pashtoons and MQMs on 31st October 1986 at Karachi and Hyderabad on the same day. When Some unknown persons opened the fire on the rally of MQM near Al-Asif Squar in Suharab Goth a Pashtoon dominated area. The clash between the Pashtoons Muhajirs reached on peak when MQM announced that the caravan of the buses went to Hyderabad from Karachi by road. After it the banners were fixed by Pashtoons at Suhrab Goth that they would not allow the Caravan of MQM to cross from their areas. When Caravan of MQM was passing from Suharab Goth on 31st October 1986, the fires were open on the procession and ethnic riots erupted once again in Karachi and Hyderabad. Many innocent Muhajirs, Pashtoons and Punjabis were killed. Punjabi officers and transporters as well as Punjabi bureaucracy fully supported Pashtoons during the riots. Jeay Sindh Tahreek supported MQM. But Palejo and other nationalist did not extend the support to MQM. After this situation many people compared Karachi with Beirut. Anees Jeelani wrote about it Karachi appears to be fast becoming another Beirut.10
7 8

K. M. Arif (2001) Khaki Shadow,, P,226 Oxford University Press Karachi www.fas.org 9 Daily Jang 26 October 1986 10 Anees Jeelani (1991) Advance to wards Democracy, The Pakistan Experience P- 273, Progressive Publishers Lahore

578

Amir Ali Chandio and Ishtiaq Ahmed Choudhry

The clash between the Pashtoons and MQM was turned into Pashtoons Punjabis and Muhajirs because during clashes between Pashtoons and Muhajirs, Punjabi fully supported Pashtoon. Ghulam Sarwar Awan leader of Punjabi Pushtoon Ithad (PPI) said that they (Punjabi) supported Pashtoons because Pashtoon leaders came to him and appealed for help and protection from the terrorism of MQM. After that they supported them.11 And the speeches of the leaders of the MQM was also the one of the reasons of the clash between the Punjabis and Muhajirs in which MQM demanded to stop the settlement of the Punjabis in Sindh province. These were the bloody ethnic riots of Sindh in which at least 50 people were killed in one night in Orangi Town, on 14th December 1986, and 70 more killed on 15th December 1986.12 It was ghastly manslaughter in Karachi. Hundred innocent Pakistani who were Punjabi, Pashtoons or Muhajir were lost their lives during the riots. According to the available statistics as many as 173 persons were killed during the riots between the Muhajirs and Punjabi and Pashtoons and at least 10 banks, 75 vehicles 24 houses and 20 shops were burnet during the last 153 days of 1986.13 During theses riots Altaf Hussain said, Karachi is no more mini-Pakistan. He further said, Sindh could not bear any more population. There are also Lahore and Faisal Abad in the way.14] The bloody riots between the Punjabis, Pashtoons and Muhajir were also continued in 1987. After winning the local body elections of 1987 MQM changed their guns towards Sindhis. Khaliq Junejo leader of Jeay Sindh Mahaz Said about it "After winning the local body elections MQM wanted to re-established their links with the establishment of Pakistan to get the privileges.15 Riots Between Sindhis and Muhajirs MQM swept the local bodies elections in the Muhajr populated cities Karachi and Hyderabad. The base of the MQMs politics was on hatred thinking. MQM opposed Punjabis and Pashtoons in its first four years. They had advocacy for the unity between the Sindhis and Muhajirs. But later in 1987, they fought Sindhis. Sindhi Muhajir riots were started through out the province; hundreds were killed and thousands were injured. The first Sindhi Muhajir riots erupted between Sindhis and Muhajirs in 1972 on language issue. The riots of 1972 immediately controlled due to the direct involvement of Zulfiqar Ali Bhutto at that time he was the head of government and he himself took keen interest in solving the dispute. The starting point of the riots of 1987 between Sindhis and Muhajirs was replacing the name of Hyder Choke into Muhajir Choke on 6th April 1987. These riots were continued till the fall of the First Nawaz Sharif government. The Sindhi Muhajr riots were more dangerous because of their thick population in the province. This is also clear from the following table.
Language Sindhi Urdu Other Total
Source:

1941 82% 19%

1951 73.8% 9.7% 16.5%

1981 52.40% 22.64% 24.96

1996 59.7% 21% 19.3%

www.apnaorg.com/test/new/article_details.php?art_id=98

The Causes of the Riots The immediate cause of the riots of 1987 was the abrogation of the name of the Hyder Bakhsk Jatoi from Hyder Choke in Gaadi Khata Hyderabad city. But actually the roots of the dispute between the Sindhis and the Muhajirs were rooted in the history of Pakistan.

11 12 13 14 15

G. Sarwar Awan Interview www.fas.org Ibid Idid Khaliq Junejo interview

Ethnic Violence in Sindh and its Repercussions on South Asia

579

In the result of the Indian partition, great Urdu speaking Muslims from India migrated to Sindh. It was the evidence of the history that people of Sindh said well come to those peoples and provided the food and shelters. The leadership of the All India Muslim League and the bureaucracy tried to settled the Muhajirs in the cities specially in Karachi and Hyderabad though it was the stand of the Sindh government that all the migrants peoples were dispersed in allover the province. But federal government opposed the views of Sindh government and dismissed the Chief Minister Muhammad Ayoub Khuhro. The Muhajir Bureaucracy occupied the key posts in federal government. So, decision went into the interests of the migrants peoples as compare to the local peoples. The evacuee property went into the hands of the Muhajirs. Many haris (farmers _ who were the farmer from the years were ousted from the lands so the contradiction from the both communities was started from those years. At that time Punjabis also supported Muhajirs. Many Sindhi medium schools were closed or changed into Urdu medium in Karachi. Therefore, thinking of Sindhis was grown against the Muhajirs. Because at that time federal government led by Liaquat Ali Khan. He was Muhajir and against the Sindhis. Its hatered thinking towards Sindhis were open when he said to Sayed Ali Akbar Shah that "Sindhis are uncivilized and Sindhi Language is the language of the illiterate persons so how could it we protected it.16 So the early years of the history of Pakistan were full with the victimizations of Sindhis. When the scheme of One Unit was launched by the Punjabi elites, Muhajirs supported them. While Sindhi opposed and protested against the scheme. They were continuing in struggle against the One Unit but Muhajir were silent. The Muhajir community extended support to religious right wing parties while Sindhis supported the liberal and secular parties such as PPP and other groups. It was proved in the general elections of 1971. After 1971 elections Pakistan divided into two parts and Bangladesh became an independent state on 16th December 1971. The rest of Pakistan handed over to Z. A. Bhutto the leader of PPP. He tried to compensate the Sindhis, which were, ignored in the previous governments. Sindhi Language Bill was passed by the majority members of the Sindh Assembly 50 members supported to the Bill only 12 members were opposed the Bill. Though it was not against the Urdu Language, but the Urdu media described it as against the Urdu language. Jamait-i- Islami and Jamait-e- Ulma-i- Pakistan (JUP) lead the Urdu speaking people and used them against the Sindhis. Many innocent Sindhis were killed Karachi. In the reactions many Urdu-speaking people were also murdered in the Sindhi dominated areas. The quota system was also the main reason of the contradiction between the Sindhis and the Muhajirs. Muhajirs demanded to end it and Sindhis favoured it. Though it was introduced in 1949 by Liaquat Ali Khan for candidates seeking entry into officer level ranks in the federal bureaucracy. Since then it has continued in one or another.17 The riots of 1972 became under control due to the direct involvement of Z.A Bhutto and an agreement was occurred between the both communities. In the language riots of 1972, Punjabis fully supported the Muhajirs because they already read Urdu as medium of education. Jamait-i- Islami arranged rally in Lahore in the favour of the Muhajirs. Punjabi Pashtoon Muhajir Mutahada Mahaz was made by Punjabi, Pashtoons and Muhajir in Sindh Province to counter the Sindhis. It was the cause of the division between the Muhajirs and Sindhis. After that Sayed (G. M.) used term Muhajir Punjabi Mustakal Mafad (Permanent interested group) and struggle against their dominance. When Bhutto ignored the advised of Sayed on national issue of Sindh. He moved against him even he tried to make alliace with Urdu speaking people against Bhutto and dominance of Punjab. The leaders of Urdu speakers were agreed with Sayed on anti line but they did not support to Sayed on his anti Punjab move. Even Sayed supported PNA movement. After the ousted of Bhutto and imposition of Martial Law the dominance of Punjab and Punjabization of military government open the eyes of Muhajirs and they established their own party
16 17

G. M. Sayed,(not mentioned) Azadi Choo Ain Chha Lai P,146 Mehtab Ali Shah,(1998) The emergence of Muhajir Quomi Movement (MQM), The Round Table P,509, October 1998, Number 348.

580

Amir Ali Chandio and Ishtiaq Ahmed Choudhry

on the basis of the Muhajir nation. Sayed supported the MQM and he only opposed the Muhajir nation, Sayed said that Muhajirs are the part of the Sindhi nation.18 The leader of MQM Altaf Hussain visited to Sayed and attended the Jalsa (Public Gathering) of Jeay Sindh Tahreek. The worker of the Jeay Sindh Tahreek protected Altaf Hussain in interior Sindh. At that time Altaf Hussain openly talked against the dominancy of Punjab and Two Nation Theory. Due to that close relations of MQM with Sindhi Nationalists, Punjabis felt threat for their interest future and made Punjabi Pashtoon Ithad (PPI) to counter the MQM in Karachi. After the riots between the Muhajir and Punjabis and the close relation of G. M. Sayed and Altaf Hussain ruling elites of Punjab felt in future so establishment was trying to create the situation in which Sindhi and Muhajirs alliance turned into enmity. For that purpose they used their agents of both sides. The worker of MQM in Hyderabad demolished the name of Hyder Bux Jatoi and riots took place between the Sindhi and Muhajirs. The Riots Aftab Ahmed Shaikh leader of MQM was elected as mayor of Hyderabad city in 1987 local bodies elections. MQM won the 1987 local elections in the Muhajir dominated urban areas of Sindh. During the Mayor ship of MQM, the name of Hyder Choke was replaced by Muhajir Choke. It was not bearable by the Sindhis, because the name of Jatoi was most respectable for the Sindhis. Jatoi dedicated his life for the rights of Sindh and peasants movement. They changed the name of Hyder Choke was the part of the policy of divide and rule of the Zia regime. After the protest of Sindhis Altaf Hussain leader of MQM directed to his workers to rewrite the name of Hyder Bakhsh Jatoi on said choke. But the anti Sindhi Lobby in MQM was active against the Sindhi. On other side Rasool Bakhsh Palejo, Qamar Bhatti, Qadir Magsi and some workers of JST also were active against the Muhajirs and they all worked according to the strategy of the establishment. Finally Sindhi Muhajirs riots spread allover the Province. Many innocent Sindhis and Muhajirs were killed. G. M. Sayed the leader of Jeay Sindh Tahreek was trying to reduce the tension between the Sindhis and Muhajirs. The same wishes of Altaf Hussain, but both leaders did not succeed to prevent the Sindhi Muhajir riots. Due to the differences in their party ranks. MQM Chairman, Azim Ahmed Tariq, Imran Farooq General secretary, Aafaque Ahmed, Badar Iqbal and Iqbal Khan (Now leaders of Muhajir Quomi Movement Haqiqi) were trying to create the quarrel with Sindhis. They supported the divided and rule strategy of Zia. In these events the agents of the establishment also played their role to instigate the politicians and common peoples of both sides. During this situation the relations became hard between President Mohammad Zia-ul- Haq and Prime Minister Mohammad Khan Junejo. Zia dismissed his government and all assemblies. Ghulam Mustafa Jatoi became caretaker Prime Minister of Pakistan. Even dismissed of government could not prevent the ethnic riots in Sindh. Hyderabad city was the center of the riots between the Muhajirs and Sindhis. On 18th June 1988 students of Sindh University were going to participate in the procession before the press club of Hyderabad, when buses of the university reached near Municipal Office, fired was opened on buses. At least 8 students were died inspite of the imposition of curfew.19 During that phenomenon unknown person attacked on Mayor Hyderabad Aftab Ahmed Shaikh on 17th July 1988. Police blamed to Bashir Qureshi and Niaz Kalani leaders of Jeay Sindh Students Federation. The riots spread in the city and at least 8 people were killed. The situation of the cities of the province became worst. G. M. Sayed told newsmen on 22nd July that both government and Altaf Hussain were responsible for the deteriorating law and order situation. Altaf Hussain had become arrogant, inflated, he said. On the following day, Altaf Hussain called on Sayed at Hyder Manzil and the two leaders made and remove misunderstandings.20
18 19 20

Khaliq Junejo Interview www.fas.org Ibid

Ethnic Violence in Sindh and its Repercussions on South Asia

581

After the dialogue between Sayed and Altaf Hussain, the tension reduced between the Jeay Sindh Tahreek and the workers of MQM but situation did not completely under control. In the month of August the APMSO students attacked on Sindhi Students in the educational institutions of Karachi. During those days General Zia died in Air crash on 17th August 1988, near Bahawalpur. Senate Chairman Ghulam Ishaq Khan became acting President of Country. General Zia was passed away but his strategy later continued. In the month of September APMSO leaders under the guidance of five councilors attacked on the Sindhi Students and lectures in the collages of Karachi and Karachi University. The riots did not control in the big cities of the province. On 30th September 1988, the worst carnage of Hyderabad history took place in which more than 150 persons. Including Urdu speaking Gujrati, Sindhi and Memon were killed.21 But K. M. Arif wrote that more than two hundred people were killed in the firing of the political militant. The majority of the killed peoples belonged to Muhajir community.22 MQM and government blamed on Jeay Sindhs faction of Qadir Magsi. On following day i.e. 1st October 1988, MQM preplanned attacked on the houses of Sindhis and more than 90 Sindhis were killed.23 In Sindhi Muhajir ethnic violence at least 1000 people of both sides were killed. 24 during the years of 1987 to 1991. In 1991 MQM splits into two groups i.e. MQM Altaf lead by Altaf Hussain and MQM Haqiqi lead by Aafaq Ahmed). After the division of the MQM the quarrel was started between them. After that the riots between Sindhis and Muhajirs were reduced. The bloody ethnic riots left wounded history of the country; the fraternity and brotherhood does not exist even still today in the land of Sindh, which was the land of peace and love, before it. Factors, which were involved in the growth of the ethnic violence of Sindh as under: Migration of the People Towards Sindh The migration of the peoples towards urban areas of Sindh province was the main reason of the ethnic tension and riots. The people migrated from the rural areas of the province to the urban areas in the search of the employment especially in Karachi. The settlement of those peoples was ultimately affected the local peoples and the growth of contradiction between them was natural. The big numbers of the peoples were migrated from the northern area of the country especially from Punjab and NWFP to Sindh. In 1981 the Census calculated a net migration to total population ratio of 9.6% for Sindh.25 The migrated peoples from Punjab and NWFP were easily succeeded to get jobs due the influence of the establishment in federal, Provincial as well as in private sectors, because the establishment of Pakistan was dominated by Punjabi and Pashtoons. The illegal migrants of Bangladesh, India, Barma and other countries were also settled in the Karachi the capital city of the Province. Many of them were absconder from their respective countries. They were involved in crime and ethnic riots. Saveda Hina quoted in her thesis that at least 20.2 million illegal migrants were settled in Karachi only and they were involved in the criminal activities most of them trained for the terrorist activities.26 The Afghan Pashtoons were fully supported to Pakistani Pashtoons and directly involved in Muhajir Pashtoons conflict. A big number of the Biharis were illegal entered in Pakistan and settled in Karachi and joined to MQM. They Biharis were trend by the religious organizations of Al-Badar and Al-Shams and they were worked for the MQM now. So Biharis also on front line during the Sindhis and Muhajir as well as Muhajir Punjabis riots.

21 22 23 24 25 26

Ibid K. M. Arif (2001) Khaki Shadows, P, 228, Oxford University Press Karachi www.fas.org Ibid Feroze Ahmed, (2000) Ethnicity, state and national integration vii: Intersection of ethnicity and class. Dawn 18 April 2000 Saveda Hina, (2003) Karachi Ka Mukadama P, 118-119, Sabih Publishers Lahore

582 Foreign Involvement

Amir Ali Chandio and Ishtiaq Ahmed Choudhry

In the ethnic riots we cannot rule out foreign involvement. Many workers of the Afghan agency KHAD were entered in Pakistan in the shape of the Mujahideen during Najeeb government. They were involved in the activities in which they created the disturbance in the cities. Numbers of Afghan Muhajirs were directivity participated in the ethnic riots. The ammunitions and weapons reached to the various ethnic groups through the Afghan people. Anees Jeelani said about the ethnic riots that Afghan policy of the government and the long years of political suppression and deviation from constitutional rule responsible for the catastrophe.27 India involved in the Sindh crises, the Indian intelligence agency Research and Analysis Wing (RAW) directly operate the activities in Pakistan to counter the Kashmir issue. The strike and disturbance created by the MQM on Kashmir Day, which was celebrated by the people of Kashmir and Pakistan on 5th February. In 1990 The government of Pakistan observed February 5 as solidarity day with the Kashmiries and rallies were held all over the country. In the evening, masked gunmen killed two persons and kidnapped eight others in an obvious bid to counter Pakistans Kashmir cause. MQM announced strike on 7th February. On 6th February 1990 the worst trouble of the year started 64 person were killed between 6th and 9th February.28 Number of Indian citizens illegally entered and settled with the help of the Urdu speaking bureaucracy and MQM. They were fully paid by the Indian government. Those migrated people were involved to create the disturbance in Sindh province in the shape of the ethno national and sectarian riots. Also some Hindus were worked for the Indian governments. Brigadier Sayed A.I Tirmazi, wrote in his book titled Profiles of Intelligence about that interference of Indian government in Sindh as under The province of Sindh seems to be on the top of their hit list RAW has been quite successful in creating a serious situation of anarchy and lawlessness in one of the most peaceful regions and the economic nerve center of Pakistan.29 Teesta Ghosh wrote India can provide moral and political support to groups such as MQM.30 Repercussions on South Asia The relations of the two states of South Asian states means Pakistan and India were not good. The main problem between them was Kashmir dispute. So both countries were trying to create the disturbance in each other territories. Therefore India supported those groups and organizations, which were in a position to create the disturbance and problems for the government of Pakistan. India also accused Pakistan to support the Kashmiri militants. So India and Pakistan were continuous interfering in each others internal problems. The ethnic riots left the impact on the relations of the neighboring countries. If any big incident took place in Sindh province the foreign department of the Pakistan said that foreign hands were involved in the incident and accused India for that. Both countries were invalid interference in the internal matters of each other and also used ethno national organizations. In this connection MQM was favourate organization for India, which was in a position to create the disturbance in the hub economic city of Pakistan. That was clear from the writings of Indian Counsel General of Karachi Mr. Mani Shanker Ayier The possibility of secession in Pakistan have decreased significantly, Because of the Soviet intervention in Afghanistan. Pakistan are more interested in that country. Wali Khan is alone and he is surrounded by the pro-establishment in Pashtoons in the NWFP. Blochs have either joined the Pakistan establishment or live in a selfimposed exile. Sindhis are supporting either PPP or other federalist parties. Only a few of them still cherish the fantasy of Sindhu Desh. And their influence in Sindh is marginal. In these circumstances
27 28 29 30

Anees Jeelani,(1991) Advance towards Democracy, The Pakistani Experience, P- 274 Progressive Publishers Lahore www.fas.org Brigadier Sayed A. I. Tirmazi, (1995) Profiles of Intelligence, P, 145, Fiction House Lahore Teesta Gosh (2003) Ethnic conflict in Sindh Ethnic conflict and secessionism in South and South East Asia causes, Dynamics, solution P, 123 Sage Publications New Delhi.

Ethnic Violence in Sindh and its Repercussions on South Asia

583

the MQM is now the unrivalled power center of Karachi politics. Though the MQM is divided against itself it is united against all non-Muhajirs. Thus the MQM is a force to be reckoned with Aiyers Pakistan Papers .31 India blamed on Pakistan for the help of the Kashmirs revolutionary and it accused to Pakistan for the activities of the Mujahideen in Kashmir. An in the counter India supported to those forces in Pakistan specially in Karachi those were in the position to create the disturbance in the city and other part of the country. H.S. Sodhi narrated it in these words The Indian objective of supporting terrorist in Pakistan, among other consideration, is aimed at pertaining the latter (Kashmir) from interfering in the internal affairs of India.32 MQM was fully supported by the Indian governments. After the challenging of Two Nation Theory by the leader of MQM Altaf Hussain, Many absconders migrated to India and they were protected and facilitated by the Indian government. So these types of the Indian action increase the tension between the neighboring states. Due to that reasons Pakistani government declared to MQM as security risk for the country Dr Mehtab Ali Sayed noted it in these words In 1992, The Pakistan Army considering the MQM as security risk, decided to flush out its anti state cadre. 33 The Indian television Door Darshan and also All India Radio service fully supported the views of the MQM, Radio and TV coverage the killing of the MQM workers without any confirmation. The simultaneously Sindhi service of All India Radio supported Sindhis in their programmes. It was the policy of the Indian government to create the disturbance for the Pakistanis rulers and get the revenge of Kashmir. On 5th February all Pakistani celebrated the Kashmir Day on that day the violence in Karachi by MQM directly weakened the case of Kashmir. This type of the action of the Indian government and the blame of the Pakistani rulers on India was created the more tension between the countries. The ethnic violation challenged the security of Pakistan and also left the impact of the regional peace. Dr Mehtab Ali Shah wrote about it as under Rising tension have threatened both the territorial integration of Pakistan and regional security arrangements of India.34 The clash between the Pashtoons and Muhajirs the Afghani Pashtoons involved in the riots and fully supported to the Pakistan Pashtoons so, the directly or indirectly Afghan government also involved in the riots. The tense relations of India Pakistan created the instability in the region special in South Asia. Both countries are nuclear power and major country of the South Asia and its relations directly or indirectly affect the entire situation of the region.

Conclusion
The ethnic violation and riots in Sindh were the result of the injustice and favoritism. The migration from internal as well as external side toward Sindh was also the main reason of the ethnic violation in Sindh province. Number of the peoples from Afghanistan, Bangladesh, Nepal, India, Iran Srilanka and other countries were illegally came and settle down here. Many of them were absconder from their respective countries and involved in the crime. The migration from the northern side of the country also created the unemployment in the capital city of the province. And it grew the contradiction and tense between the ethnic groups. The foreign involvements also grew those riots in the province. The Indian government supported the both groups, which were in a position to create the disturbance in the country, and
31 32

33
34

Mehtab Ali Shah,(1998) Emergence of Muhajir Quami Movement(MQM) in Pakistan and implications for Regional Security, The Round Table No: 348, October 1998, Carfex Publishing Ltd. H. S. Sodhi,(1993) India internal trouble and foreign factor, Indian Defence Review, Delhi, Vol> 8, No 2 April 1993, P, 83. Mehtab Ali Shah,(1998) Emergence of MQM., Round Table, P, 512. Mehtab Ali Shah,(1997) Ethnic Tension in Sindh and their possible solution, Contemporary South Asia, Volume 6 Number 3,1997 P, 260

584

Amir Ali Chandio and Ishtiaq Ahmed Choudhry

though that tactics India counter the Kashmir issue in Sindh. The Indian electronic as well as print media played vital role to encourage those groups. The Indian media described those groups as suppressive and neglected groups. At a same time All India radio Sindhi service supported to the Sindhis and All India radio Urdu service supported the Urdu speaking Muhajirs. Afghan government also involved in the ethnic tension in Sindh Afghan Pashtoons fully supported Pakistanis Pashtoons and in the Orangi crises Afghan were directly involved. Due to those policies of the neighboring countries the relations became in tense and the impact on the south Asia politics. And the Asian peace and security would in threat the Kargil crises also linked with those policies of the neighboring countries.

Suggestion
1) 2) 3) 4) 5) 6) The both countries may refrain from the interference in the internal matters of the respective countries. The neighboring countries would be avoids using media against each other. The respective countries may avoid supporting any ethnic groups. The respective countries would try to solve disputes through the dialogues and negotiation. The countries would sign an accord that they would not to protect any criminal. And also sign the treaty to hand over the criminal to each other. Through these measures it would be fruit full for the peace and security and also development for the region.

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.585-591 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Influence of Supervisors Gender on Mathematics Student Teachers Performance in Teaching Practice


Ugboduma, S. O Mathematics Department Delta State College of Sports and Science Education, Mosogar Nigeria E-mail: Ugbosam(2007)@yahoo.Com Tel: 08027905667 EZE, A.E. Mathematics Enugu State University of Science and Technology Enugu Abstract This paper examined the influence of supervisors gender on mathematics student teachers performance in teaching practice. Four hypotheses were formulated for the study, and t-test statistics was used to test the hypotheses. The study was limited to students of mathematics department of Faculty of Education, Delta State University, Abraka. The research method used was ex-post facto design and the instrument use was direct observation and official records which consisted of teaching practice results for 2003/2004 and 2004/2005 Sessions. Two research questions were adopted for the study. Analysis of results revealed the following findings: Male students performed higher with average mean of 3.78 in teaching practice than their female counterparts with average mean of 3.62, mean difference of 0.16 in mathematics department despite the supervisors; and student supervised by male supervisors performed higher with average mean of 7.83 in teaching practice than those supervised by female supervisors with average mean of 6.96, mean difference of 0.87. The difference observed was however not statistically significant showing that supervisors gender does not influence mathematics student teachers performance in teaching practice.

Introduction
The professional education of teachers cannot be valid without adequate preparation of prospective teachers for dynamic roles in todays schools through teaching practice. Oluremi (1990) in the Teachers Forum Conference noted A few years ago, it was the scene in Nigeria to find only privileged children in School because of financial constraints upon many parents. Now, the Nigeria Educational System has increased tremendously for children of different background, aptitude, behaviours, and in learning ability. The Universal Basic Education (UBE) has also ushered in many children from upper, middle and lower classes into the school system. All these, therefore, call for a more thorough preparation of future teachers. The National Policy on Education (FRN, 2004) dwells on the objectives of teacher education as: a) To produce highly motivated, conscientious and efficient classroom teachers for all levels of our educational system. b) To encourage further the spirit of enquiry and creativity in teachers; c) To enhance teachers commitment to the teaching profession.

Influence of Supervisors Gender on Mathematics Student Teachers Performance in Teaching Practice d)

586

To provide teachers with the intellectual and professional background adequate for their assignment and make them adaptable to changing situations. From the above, the teacher education programme is aimed at turning out effective professional teachers. The prospective teachers should avail themselves of the opportunity provided by the programme as a means of testing their suitability for the profession. Jeffreys (1991) treated with disdain the training component of the preparation of the teacher. He went on to state that it is more important in training teachers and to produce well-educated people than to produce technically competent practitioners. For student teachers to be competent in teaching, they need real practical orientation by way of teaching practice (Stones and Morris, 1993). The teaching competency is synonymous with intellectual curiosity; reasoned judgment and action-based on creative and reflective thinking; attitude of human acceptance and aesthetic sensitivity; and understanding of the fundamental concepts, principles and ways of thought of the teaching profession. Davis (2001) sees teaching practice as entangled in a mass of confusion unmade decisions, and experiences without a comprehensive definition and clear-cut statement of goals and purposes. Stones and Morris (1993); generalized teaching practice as all the learning experience of student teachers in schools. They hold the following concept as ambiguous because: a) Of the practice of teaching skills and the acquisition of the role of a teacher; b) Of the whole range of experiences that students go through in schools, and c) The practical aspects of the course are distinct from theoretical studies. Teaching practice is one of the integral parts of assessing performance of student teachers competency in classroom management and control. Money (1990) and Okorie (1999) call for effective supervision and evaluation of students on teaching practice if so as to maintain standard, recognizing that the practice period is short compared to other disciplines. Poor performance in teaching practice hinders the effectiveness and most relevant learning experience or activity to which pre-service teacher must be exposed. This shows that meaningful learning activities represent the heart of the curriculum because they are influential in shaping the learners experience and his education. Zias (1989), opined that Good intentions, fine goals, and objectives, excellent content, flawless evaluation procedures which determine the performance of students (males and females), then, are all for naught if the learning activities in which students engage do not provide them with experience whose consequences are educational. Yearly, Delta State University, Faculty of Education, Abraka mathematics education undergraduate students go for teaching practice in secondary schools where they interact with new people, new problems and new situations and react to them, with the help of teachers. Lecturers are assigned to supervise these student teachers in order to enhance effectiveness in the field of teaching and determine how competent they are in teaching; Many students have complained that supervisors contributed to their failures in mathematics teaching practice. To what extent is the statement true? This is what this study set out to examine the influence supervisors gender has on mathematics student teachers performance in teaching practice.

Problem of the Study


Despite the importance accorded teaching practice in schools, there exists a large number of student teachers failures in mathematics. This does not augur well for the nations desire for sound teachers training education development. Base on this, the study was determined to examine the influence of supervisors gender on mathematics student teachers performance in teaching practice. Significance of the study The study will create awareness in supervisors on how best to evaluate student teachers in class during supervision without bias. It will also help to check the influence of scoring student teachers by supervisors gender and having regular update of students evaluation scores.

587

Ugboduma, S. O and EZE, A.E

Research questions
The study adopted the following research questions: (1) What is the performance level of mathematics student teachers supervised by male supervisors in teaching practice in mathematics department of Delta State University, Faculty of Education, Abraka? (2) What is the performance level of mathematics student teachers supervised by female supervisors in teaching practice in mathematics department of Delta State University, Faculty of Education, Abraka? Hypotheses The following hypotheses were stated and tested at 0.05 alpha level of significance HO1: There is no significance difference in the performance of mathematics student supervised by male supervisors in 2003/2004 session teaching practice. HO2: There is no significance difference in the performance of mathematics student supervised by female supervisors in 2003/2004 session teaching practice. HO3: There is no significance difference in the performance of mathematics student supervised by male supervisors in 2004/2005 session teaching practice. HO4: There is no significance difference in the performance of mathematics student supervised by female supervisors in 2004/2005 session teaching practice. Methods This is expost facto type of research. This study was conducted using teaching practice results of mathematics student teachers (males and females) of mathematics department, Faculty of Education, Delta State University, Abraka. Students with various grades supervised by male and female supervisors were randomly selected from the mathematics teaching practice students population of 125 (2003/2004 session) and 112 (2004/2005 session). These gave a total population of 237. Using proportional random sampling, a total of 160 students were selected for the study (40 students supervised by male supervisors and 40 students supervised by female supervisors) in 2003/2004 and 2004/2005 sessions respectively. The results of the supervisors scores of the mathematics student teachers teaching practice performance were analysed using the t-test to find any significant. The decision rule for the hypotheses was that when the calculated t value at 0.05 level of significance and the appropriate degree of freedom was equal to or more than the table t-value, the null hypothesis stated was rejected; but when the calculated t-value was less than the table t-value, at 0.05 level of significance and at an appropriate degree of freedom, the hypothesis of no difference was retained. teachers teachers teachers teachers

Results
HO1: There is no significance difference in the performance of mathematics student teachers supervised by male supervisors in 2003/2004 session teaching practice.
Table 1: T-test results for mathematics teachers supervised by male supervisors in 2003/2004 teaching practice
No of students 40 40 Mean 4.15 3.98 S.D 3.58 3.84 Cal. T-value 0.205 Critical t-value 1.98 Df 78 Mean diff. 0.17 Remark NS

Groups Males Females

The obtained t-calculated value of 0.205 was less than the t-critical value of 1.98 at df = 78. Thus, the hypothesis is retrained. The result reveals that at 0.05, there is no significance difference in

Influence of Supervisors Gender on Mathematics Student Teachers Performance in Teaching Practice

588

the performance of mathematics student teachers supervised by male supervisors in 2003/2004 session teaching practice. HO2: There is no significant difference in the performance of mathematics student teachers supervised by female supervisors in 2003/2004 session teaching practice.
Table 2: T-test result for mathematics student teachers supervised by female supervisors in 2003/2004 teaching practice.
No of students 40 40 Mean 3.23 3.60 S.D 2.97 3.36 Cal. T-value 0.522 Critical t-value 1.98 Df 78 Mean diff. 0.37 Remark NS

Groups Males Females

The t-calculated value of 0.522 was found to be less than the t-critical value of 1.98, df=78. Therefore, the hypothesis was not rejected. The result, show that at 0.05, there is no significant difference in the performance of mathematics student teachers supervised by female supervisors in 2003/2004 session teaching practice. HO3: There is no significant difference in the performance of mathematics student teachers supervised by male supervisors in 2004/2005 session teaching practice.
Table 3: T-test result for mathematics student teachers supervised by male supervisors in 2004/2005 teaching practice
No of students 40 40 Mean 4.25 3.28 S.D 4.17 3.21 Cal. T-value 1.17 Critical t-value 1.98 Df 78 Mean diff. 0.97 Remark NS

Groups Males Females

The obtained t-calculated value of 1.17 was less than the t-critical value of 1.98 at df=78. Thus, the hypothesis was retained. The results on table 3 at 0.05 show that there is no significant difference in the performance of mathematics student teachers supervised by male supervisors in 2004/2005 session teaching practice. HO4: There is no significant difference in the performance of mathematics student teachers supervised by female supervisors in 2004/2005 session teaching practice.
Table 4: T-test result for mathematics student teachers supervised by female supervisors in 2004/2005 teaching practice.
No of students 40 40 Mean 3.48 3.60 S.D 3.26 2.38 Cal. T-value 0.188 Critical t-value 1.98 Df 78 Mean diff. 0.12 Remark NS

Groups Males Females

The t-calculated value of 0.188 was less than the t-critical value of 1.98. Therefore, hypothesis was not rejected. Results of table 4 at 0.05 shows that there is no significant difference in the performance of mathematics student teachers supervised by female supervisors in 2004/2005 session teaching practice.
Table 5a: Paired sample t-test result for mathematics student teachers supervised by male supervisors in both 2003/2004 and 2004/2005 sessions teaching practice.
Groups Males Females No of students 80 80 Mean 4.20 3.63 S.D 6.05 4.95 Cal. T-value 0.624 Critical t-value 1.96 Df 158 Mean diff. 0.57 Remark NS

589

Ugboduma, S. O and EZE, A.E

It is observed in the table that the t-calculated value of 0.624 was less than the t-critical value of 1.96 at df of 158. This implies that male supervisors did not influence mathematics student teachers performance in teaching practice and there was no significant difference.
Table 5b: Paired sample t-test result for mathematics student teachers supervised by female supervisors in both 2003/2004 and 2004/2005 sessions teaching practice.
Groups Males Females No of students 80 80 Mean 3.35 3.60 S.D 4.56 4.60 Cal. T-value 0.345 Critical t-value 1.96 Df 158 Mean diff. 0.25 Remark NS

The t-calculated value obtained was 0.345, less than the t-critical value of 1.96 at df of 158. This means that female supervisors does not influence mathematics student teachers performance in teaching practice. Table 5b; reveals that the paired sample t-test for mathematics student teachers supervised by female supervisors in both 2003/2004 and 2004/2005 sessions teaching practice have no significant difference.

Discussion
The results from the tables above, revealed that mathematics student teachers performance was not influence by supervisors gender in Delta State University, Abraka, The results further show that students supervised by male supervisors in both sessions (2003/2004 and 2004/2005) performed slightly high with average mean of 7.83 than those supervised by female supervisors with 6.96 of mean difference of 0.87. The difference observed was however not statistically significant. This is in consonance with Lantz (1997) studies that significant variation in the ratings of the same student behaviour can result from personal coloration and interpretation of his behaviour. This also agrees with Ezeudu (1995) findings that there is no significant difference in the mean interest scores of boys and girls in algebra despite the teachers gender. Gbodi and Laleye (2006) findings revealed that gender had no effect on achievement mean scores of students. Ajiboye (1996); Hyde (1981) in Gbodi and Laleye (2006) conducted research findings were in consonance with this findings that, there is no significant main effect of sex on students achievement in disciplines such as science, social studies and geography respectively. However, the results disagreed with Alonge and Ojerinde (1987); Chopping (1975) and Money (1990) findings that there was significant difference in the achievement of male and female students, with the male students performing better despite the supervisors. From the findings, the study concludes that mathematics student teachers performance does not in any way influence by supervisors gender in mathematics department teaching practice, Faculty of Education, Delta State University, Abraka. Recommendations The following recommendations are made based on the findings: All secondary schools in Nigeria should be fully equipped with all necessary mathematics materials so that teachers, both permanent and student teachers, can make maximum use of them in teaching at all times. Mathematics department of institutions should adopt stimulation technique during their methodology classes. In stimulation, student teachers teach their peers in micro-lessons, followed by discussion and lesson critique. Stones and Morris (1993), supported stimulation techniques by putting students into the classroom to learn on their own or lecturing them about the classroom and how to teach. It helps to remove the risk and stage fright from the trainee teacher.

Influence of Supervisors Gender on Mathematics Student Teachers Performance in Teaching Practice

590

Supervisors appointment should be the sole responsibility of the head of mathematics department. This will enhance specialist supervisors for mathematics teaching by the student teachers. Tertiary institutions should provide vehicles and remuneration for supervisors. This will make them show concern for the exercise and supervise effectively without being in haste. Without good remuneration there will be no effective service. There is need to alert the supervisors about the human pitfall to the artifacts of attractiveness, expectancy effect and so on. Assessment decisions must be base on objective criteria, spend enough time observing lesson of the students being assessed and holding a Post Assessment Conference (P.A.C.) with the student teachers in order to clarify doubts and issues. Student teachers should make use of relevant teaching aids to enhance effective mathematics concept teaching. The author holds the same opinion with Mkpa (1989) that teaching aids increase the rate of learning, facilitate retention of what is learned and hence enhances performance.

591

Ugboduma, S. O and EZE, A.E Alonge, M & Ojerinde, A. (1987). Guessing tendency and differential performance of secondary school students on Mathematics achievement test. Journal of Science Teachers Association of Nigeria (JSTAN) 25(2), 30-38 Chopping, B. (1975). The correlation for guessing in objective test. Stockholm . IEA. Davis, D. (2001). Student teaching. In R. L. Ebel (Ed) Encyclopaedia of Education Research 76-86. London: Macmillan. Ezeudu, F. O. (1995). The effects of concept mapping on students achievement interest and retention in organic chemistry. Unpublished Ph.D. thesis, University of Nigeria. Nsukka. Federal Republic of Nigeria (FRN) (2004).National Policy on Education (4th Edition). Lagos: NERDC Press. Gbodi, E. B. & Laleye, A. M. (2006). Effect of videotape instruction in learning Integrated Science. Journal of Research in Curriculum and Teaching (JRCT) 1 (1), 10-19. Jefffreys, M. V. C. (1991). Revolution in teacher education (4th Edition). London: Pitman. Lantz, D. L. (1997). The relationship of university supervisors and supervising teachers ratings to observe student teacher behaviour. Journal of American Educational Research. Mkpa, M. A. (1989). Status of implementation of the National Policy as it relates to Instructional materials. Nigeria Educational Research Association Journal 1 (1), 253-258. Money, F. O. (1990). Towards a more effective supervision and evaluation of student teachers during teaching practice. In the seminar series paper on teaching practice presented to Faculty of Education, University of Benin, Benin City: Ilupeju Press Ltd. P. 39, 40. Okorie, J. U. (1999). Fundamentals of teaching practice (2nd Edition). Enugu: Forth Dimension Publishers Ltd. Oluremi, O. J. (1990). Teacher forum of 1990, entitled teaching practice. Stones, E. & Morris, S. (1993). Teaching practice: Problems and perspective (3rd Edition).London: Methuen and Company Limited. Zias, R. S. (1989). Curriculum, principles and foundations (2nd Edition). New York: Thomas. Y. Crowell Company Inc.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.592-604 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) : Exemple des Commune DAbobo, Adjam, Attcoub, Cocody et Port-Bout
Aman Messou Laboratoire dEnvironnement et de Biologie Aquatique UFR - Sciences et Gestion de LEnvironnement, Universit dAbobo-Adjam 02 BP 801 Abidjan 02, Cte dIvoire Lacina Coulibaly 02 BP 801 Abidjan 02, Cte dIvoire E-mail: coulacina2003@yahoo.fr Tel : 00225-07-49-71-53 Allassane Ouattara Laboratoire dEnvironnement et de Biologie Aquatique UFR - Sciences et Gestion de LEnvironnement, Universit dAbobo-Adjam 02 BP 801 Abidjan 02, Cte dIvoire Germain Gourne Laboratoire dEnvironnement et de Biologie Aquatique UFR - Sciences et Gestion de LEnvironnement, Universit dAbobo-Adjam 02 BP 801 Abidjan 02, Cte dIvoire Abstract La contribution des garages automobiles itinrants dans la pollution de lenvironnement dAbidjan a t value par une enqute auprs des garagistes dAbobo, Adjam, Attcoub, Cocody et Port-Bout. Des chantillons de sols ont t prlevs dans certains garages dAdjam pour apprcier le niveau de pollution du profil des sols en fonction du temps. Au total, 300 garages ont t enquts, pour un effectif de 5 363 personnes dont 4 231 apprentis et 1 132 chefs de section. Ces garages occupent 16.6 ha, dont 12.2 ha de surface active et 4.4 ha de surface morte. Les dchets gnrs par les activits des garagistes ne sont pas collects. Par ailleurs, ils sont abandonns sur les surfaces actives qui ne sont pas amnages (98,6 % de sols nus) et contribuent la pollution de leurs sols et de lenvironnement extrieurs. En effet, la conductivit et les matires sches volatiles de la couche suprieure des sols des surfaces actives sont significativement suprieures (p<0.05) couches des sols des surfaces mortes. Par ailleurs, les sols des surfaces actives sont acidifis par rapport aux sols des surfaces mortes. Les garages automobiles itinrants constituent donc des sources de pollution des sols et lenvironnement extrieur du district dAbidjan.

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) : Keywords: Garage automobile, Abidjan, environnement, pollution, sol, dchet.

593

Introduction
La pollution de lenvironnement par les activits industrielles et artisanales est bien connue travers le monde. Dans les pays en dveloppement, la caractrisation des sources de pollution de lenvironnement, notamment par les activits de rparation des vhicules nest pas effectue. Cependant, celles-ci gnrent des polluants (huiles de vidange, plomb) qui ont un caractre nocif avrs pour lenvironnement. Par exemple, les huiles de vidange contiennent des mtaux (As, Cd, Cr, Ni, Pb, etc.) et des polluants organiques (phnanthrne, fluoranthne, pyrne, benzne, etc.) dont certains sont mutagnes ou cancrignes (Haupt et al., 1996; Surprenant et al., 1983; Environnement Canada, 1993; Alloway et Ayres, 1997). Par ailleurs, ces huiles ont un impact ngatif sur la faune et la flore aquatique (Bate et Crafford, 1985 ; Hedtke et Puglisi). Pour Grimmer et al. (1981), la mauvaise gestion des huiles de vidange constitue tout simplement une menace pour la population humaine et les cosystmes. Quant Cairncross et al. (2004), ils estiment que les problmes lis lenvironnement sont la cause de 21 % des maladies dans le monde, mettant ainsi clairement en vidence une relation entre la sant humaine et celle de lenvironnement. La ville dAbidjan est le sige dimportantes activits artisanales dont les dchets sont mal grs. Ces dchets polluent lenvironnement macroscopique (physique) et microscopique (sol, eau et air). Par exemple, les dchets des garages automobiles itinrants ne sont pas grs comme au Qubec o ils sont recycls (Environnement Canada, 1999). Dans le cadre de llaboration dune politique de gestion durable de lenvironnement urbain dAbidjan, limpact des garages automobiles itinrants doit tre apprhend. Ainsi, la prsente recherche se propose-t-elle dapprcier limpact des activits des garages itinrants de rparation des vhicules sur lenvironnement. Les garages automobiles itinrants seront caractriss et le diagnostic de la gestion des dchets des garages permettra dvaluer leur contribution dans la pollution de lenvironnement physique des communes dAbobo, Adjam, Attcoub, Cocody et de Port-Bout.

Materiel et Methodes
Zone d'tude Situ au sud-est de la Cte dIvoire, la ville dAbidjan (5 N et 4 W) stend sur environ 58 000 ha dont 85% de terre ferme et 15% de lagune. Abidjan est compos de 10 communes dont celles dAbobo, Adjam, Attcoub, Cocody et de Port-Bout (Figure 1) qui font lobjet de cette tude. Son climat, de type quatorial de transition est caractris par quatre saisons dont deux saisons sches (grande saison sche de dcembre mars et une petite saison sche daot septembre) et deux saisons de pluies (grande saison des pluies davril juillet et une petite saison des pluies doctobre novembre) (Climatezone.com, 2006).

594

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne


Figure 1: Communes chantillonnes dans le District dAbidjan.

Communes chantillonnes

Evaluation de la pollution solide et liquide Une enqute a t effectue auprs des garages automobiles itinrants des communes dAbobo, Adjam, Attcoub, Cocody et de Port-Bout, afin de rcolter des informations relatives lge des garages, la production ainsi que la gestion des dchets gnrs par les activits de rparation de vhicules. Elle a t faite par les techniques d'observation et d'administration de questionnaire. Ce sont les garages situs le long des grandes voies de circulation routire ainsi que ceux qui sont accessibles qui ont t chantillonns. Les quantits de dchets solides ont t calcules partir des donnes de rfrences de la composition dun vhicule de lAgence de lEnvironnement et de la Matrise de lEnergie (ADEME) (2000). Analyse des paramtres physico-chimiques des sols Les chantillons de sols ont t prlevs dans 8 garages dans la commune dAdjam. Les garages ont t retenus sur base de leur anne de cration entre 1970 et 2004. Les chantillons ont t prlevs sur les surfaces mortes (surface de stockage sans activit de rparation) qui ont t prises comme surface tmoin et les surfaces actives (espace de droulement des rparations des vhicules). Les surfaces des garages sont gnralement de type rectangulaire. Les chantillons de sol ont t prlevs sur les sommets et le centre du plus grand rectangle pouvant occuper lespace disponible sur les surfaces actives ou mortes (Figure 2). Cette mthode dchantillonnage a t adopte pour prendre en compte les disparits de la rpartition des polluants sur les deux types de surfaces. Lchantillonnage a t effectu laide dune tarire dans les couches suprieures ([-10, 0 cm]) et infrieures ([-20, -10 cm]) (Figure 2A). Dans chaque garage chantillonn, lensemble des 5 points du rectangle a constitu pour une couche donne (suprieur : E1, E2, E3 E4 et E5; infrieur : E1, E2, E3 E4 et E5), lchantillon de ladite couche. Les chantillons ont ensuite t conservs dans des sachets en plastique pralablement annots. Au laboratoire, chaque chantillon a t pass au tamis AFNOR de maille 2 mm de diamtre et le passant a t rcupr pour les diffrentes analyses.

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) :

595

Figure 2: Echantillonnage des sols sur le rectangle tabli sur lespace disponible sur les diffrentes surfaces (actives ou mortes) des garages. A : Vue plane de la surface suprieure de chaque couche (trait plein :[-10, 0 cm] ; trait discontinu : [-20, -10 cm]). B : Vue en perspective des deux couches chantillonnes (trait plein :[-10, 0 cm] ; trait discontinu : [-20, -10 cm]) sur les diffrentes surfaces.
E4 E5 Couche suprieure Profondeur (cm) E3 E4 E5 E3 A E2 E2 E1 E1 Niveau du sol E3

E4

E5

E2

E1

-10

E4 E3

E5

E1 E2

Couche infrieure -20 B

Les matires grasses ont t extraites l'eau distille chauffe 100C. Dix (10) grammes dchantillon de sol ont t pess et disposs dans un bcher. Ensuite, 25 ml d'eau distille 100C ont t ajouts dans le bcher et lensemble a t agit sur une plaque agitante barreau magntique 150 rotation par minute (rpm) pendant 30 minutes. La suspension a t dcante pendant 10 minutes et le surnageant a t recueillie dans un creuset vide (masse : MCV) et sches ltuve 105C pendant deux heures. La masse de matires grasses (MMG) a t obtenue en faisant la diffrence entre la masse du creuset vide et sa masse avec l'extrait sch (MCE) suivant la relation (1). MMG = MCV - MCE (1) Lefficacit de lextraction des matires grasses leau chaude a t tudie sur du sol (10 g) enrichi en huile de vidange (0,0625 2 g). Les matires sches volatiles (MSV) ont t obtenues en faisant la diffrence entre la masse dun chantillon de sol sch 105C (M1) et sa masse aprs combustion pendant deux heures 525C (M2) (Garay et al., 1986) suivant la relation (2). MSV = M1 - M2 (2) Laccumulation des MSV sur les surfaces actives des garages a t dtermine en faisant la diffrence entre la masse de MSV dune couche donne ([-10, 0cm] ou [-20, -10 cm]) de sol de la surface active avec celle de la couche quivalente sur la surface morte. Le pH des chantillons de sols a t mesur laide dun pH-mtre pH330/SET (Danida, Danemark). Vingt (20) g dchantillon de sol ont t placs dans un bcher et on y a ajout 50 ml deau distille dsionise. Lensemble a t agit laide dun agitateur magntique 150 rpm pendant 10 min et on a laiss dcanter la suspension pendant 30 min. Ensuite, la sonde de llectrode pH a t introduite dans la solution pour mesurer le pH. La conductivit des chantillons de sols a t mesure dans une solution de sol prpare comme lors de la mesure du pH. Lensemble a t agit 150 rpm pendant 10 min et laiss au repos pendant une heure. Ensuite, la conductivit du surnageant a t mesure laide dun conductimtre de type LF 197 WTW.

596 Analyses statistiques

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne

Les donnes physico-chimiques recueillies (pH, Conductivit, des Matires grasses et MSV) dans les couches des diffrentes surfaces ont donn lieu des analyses de variance (ANOVA) pour vrifier la signification des diffrences ou similarits entre diffrents chantillons. Lanalyse est effectue laide du programme statistica 7.01 qui donne la valeur de p (p-level). Si une valeur de p est infrieure 0.05 la diffrence est significative. Dans le cas contraire elle ne lest pas.

Resultats
Les garages automobiles itinrants enquts sont au nombre de trois cent (300), repartis dans les secteurs de la mcanique, llectricit, la peinture et la tlerie. On y dnombre 5 363 mcaniciens, dont 4 231 apprentis et 1 132 chefs de section. Il y a une augmentation progressive du nombre de garages de 1964 2004 (Figure 3). Leur regroupement par tranche d'ge de 5 ans permet de constater que la classe [1995, 2000] est la plus importante (97 garages). Elle est suivie par la classe [2000-2004] qui renferme 82 garages. En revanche, les classes [1965, 1970] et [1970, 1975] sont les moins importantes (3 garages). La superficie totale occupe par les garages est 16,6 ha dont 12,2 ha de surface active et 4,4 ha de surface morte. Les surfaces actives sont constitues 98,7 % de sols nus, 1 % de surface btonne et 0,3 % de surface goudronne. Les garages sont quips 85 % de lunettes de protection, 18 % de combinaisons, 4,6 % de gants et 0,3 % de chaussures spciales.
Figure 3: Structure par classe dge des garages automobiles dans les communes dAbobo, Adjam, Attcoub, Cocody et Port-bouet en fonction de leur priode de cration.
120 100

N om bre de garage

80 60 40 20 0 [1965 - 1970[ [1970 - 1975[ [1975 - 1980[ [1980 - 1985[ [1985 -1990[ [1990 - 1995[ [1995 - 2000[ [2000 - 2004]

Classes d'ge

Les dchets gnrs dans les garages se composent de solides et de liquides (Tableau 1). Les carcasses (575,9 t) et les vhicules stationns de plus dun an (632,7 t) dominent la composition des dchets solides. Ensuite, suivent les plastiques (126,5 t) et les pneus uss (85 t). Ces dchets sont souvent intgrs aux dchets municipaux ou rejets dans la nature (Figure 4A). Quant aux dchets liquides, ceux-ci se composent d'eaux uses (710,7 m3/mois), d'huile de vidange (28,2 m3/mois) et de carburants uss (2,9 m3/mois). Les huiles de vidange sont collectes dans diffrents rcipients (bidons, carters, fts, fosses), dverses sur le sol, dans les canalisations deau pluviale et dans des fosses (Figure 4B). Cependant, une partie des carcasses, huiles de vidange, batteries et pneus uss est valorise par les garagistes et les artisans (Figure 5). Concernant les huiles de vidange, dans 81% des

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) :

597

garages, les mcaniciens les utilisent pour le graissage des pices et dans 33% de garages, les artisans les utilisent pour lalimentation des machines de carrires. Quant aux carcasses de vhicules, dans 66% des garages, les artisans les transforment en divers ustensiles (fourneaux, pelles, cuvettes, arrosoirs, dabas, etc.). Cependant, ces carcasses sont utilises dans 56% des garages par les mcaniciens pour substituer certaines parties de vhicules endommages. Le plomb des batteries uses est utilis par les mcaniciens dans 27% de garages pour faire la soudure et dans 87% de garages par les artisans pour constituer des lests pour les engins de pche. Concernant les pneus uss, ils sont utiliss comme pneus de charrettes par les mcaniciens dans 57% de garages ou incinrs par les artisans dans 77% de garages pour rcuprer les fils mtalliques qui servent la fabrication de grilles de fourneau. Il faut relever que lincinration anarchique des pneus contribue la pollution de lair. Quant aux carburants uss, ils sont essentiellement utiliss dans 91% de garages par les mcaniciens comme carburant dans dautres vhicules ou comme solvant de dgraissage de pices automobiles.
Figure 4: Pollution de lenvironnement physique par les dchets des garages. A : carcasses de vhicules jetes sur une rive de la lagune Ebri Abattoir, Port-Bout. B : fosse de rception dhuile de vidange dans un garage itinrant Abobo-Anador, Abobo.

Figure 5: Niveau de valorisation des dchets gnrs dans les garages automobiles itinrants.
1 Ratio garages Ratio artisans

0,75

Ratio

0,5

0,25

0
Carcasses Carburants uss Batteries uses Pneus uss Huiles de vidange

Nature des dchets

598
Table 1:

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne


Dchets solides et liquides gnrs dans les garages automobiles itinrants des communes dAbobo, Adjam, Attcoub, Cocody et Port-bouet.
Nature des dchets Pneus Carcasses Vhicules dge > 1an Plastiques Vitres Batteries Huiles de vidange Eaux uses Hydrocarbures uss Quantit totale 85,1 tonnes 575,9 tonnes 632,7 tonnes 126,5 tonnes 2,6 tonnes 34,6 tonnes 28,2 m3/ mois 701,7 m3/ mois 2,9 m3/ mois

Type de dchets

Solides

Liquides

La Figure 6 prsente lvolution du pH dans les horizons des sols des surfaces actives et ceux des surfaces mortes. Sur les surfaces mortes, le pH de la couche suprieure est lgrement suprieur celui de la couche infrieure pour tous les garages lexception de ceux crs en 1980 et 1984 (Figure 6A). Sur les surfaces actives, le pH de la couche suprieure est suprieur celui de la couche infrieure (Figure 6B). Les pH des horizons de la surface active des garages crs en 1970 sont les plus levs (couche suprieure : pH = 8 ; couche infrieure : pH = 7,9). Par ailleurs, au niveau des surfaces actives, ce sont les garages crs en 2002 qui ont les pH les plus bas (couche suprieure : pH = 5.9 ; couche infrieure : pH = 5,7). Cependant, lanalyse de variance ralise sur les pH des sols des surfaces actives et ceux des surfaces mortes na pas donn de diffrence significative entre les diffrentes surfaces (p > 0,05) ainsi quentre les horizons dune mme surface (p > 0,05).
Figure 6: Variation du pH dans les profils ([-10, 0 cm] et [-20, -10 cm]) des surfaces active et morte des sols des garages automobiles itinrants en fonction de l'ge des garages. A : Surface morte ; B : Surface active
2002 Age des garages 'anne) 2000 1995 1987 1984 1980 1975 1970 0 2 4 pH 6 8 10
]-20, -10 cm] S morte ]-10, 0 cm] S morte

2002 Age des garages (anne) 2000 1995 1987 1984 1980 1975 1970 0 2 4 pH 6

]-20, -10 cm] S active ]-10, 0 cm] S active

10

La Figure 7 prsente lvolution de la conductivit dans les horizons des sols des surfaces actives et ceux des surfaces mortes. Sur les surfaces mortes, la conductivit des couches suprieures (82EC163 S/cm) est globalement suprieure celle des couches infrieures (55EC124 S/cm) (Figure 7A). Sur les surfaces actives, on note le mme profil dvolution de la conductivit dans les horizons des sols. Cependant, la conductivit de la couche suprieure des surfaces actives (95EC656 S/cm) est suprieure celle de la couche suprieure des surfaces mortes. De mme, la conductivit de la couche infrieure des sols des surfaces actives (89EC506 S/cm) est suprieure celle des surfaces mortes. Les conductivits les plus leves sur les surfaces actives se rencontrent dans les couches des garages crs en 1980 (656 S/cm) et 1970 (508 S/cm). Lanalyse de variance effectue

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) :

599

sur les conductivits des diffrentes surfaces des garages a donn une diffrence significative entre les deux couches des surfaces actives et celles des surfaces mortes (p<0.05). Cependant, la diffrence ntait pas significative entre les deux couches dune mme surface (morte ou active) (p = 0.67).
Figure 7: Variation de la conductivit dans les profils ([-10, 0 cm] et [-20, -10 cm]) des surfaces active et morte des sols des garages automobiles itinrants en fonction de l'ge des garages. A : Surface morte ; B : Surface active
A
Priode de cration des garages

2002 Priode de cration des garages 2000 1995 1987 1984 1980 1975 1970 0 200 400 600

2002 2000 1995 1987 1984 1980 1975 1970


800

]-20, -10 cm] ]-10, 0cm]

]-20, -10 cm] ]-10, 0 cm]

200

400

600

800

Conductivit lectrique (S/cm )

Condcuctivit lectrique (S/cm )

Lefficacit dextraction des matires grasses des sols a t tudie et est prsente en annexe. Pour des concentrations de matires grasses dans les sols comprises entre 0,00625 et 0,0125 g g-1, la mthode permet dextraire entre 94 et 97 % de celles-ci. Au del de 0.0125 gg-1 de matires grasses dans les sols, leur rendement dextraction chute pour atteindre 40% 2 gg-1. La Figure 8 prsente laccumulation des matires grasses dans les horizons des sols des surfaces actives des garages. Celleci est globalement plus importante dans la couche suprieure (1987 : 0,024 g ; 1980 : 0,054 g) par rapport la couche infrieure (1970 : 0,008 g ; 1980 : 0,015 g). Par contre, cest dans les garages crs en 1995 quaucune accumulation de matires grasses (-0,004 g) na pu tre observe dans les horizons des sols. Lanalyse de variance effectue sur les matires grasses accumules dans les horizons des sols a donn une diffrence significative entre les couches suprieures des surfaces actives et celles des surfaces mortes (p < 0,05) et trs significative avec celles des couches infrieures des surfaces mortes (p < 0,005). Sur une mme surface, la diffrence nest pas significative entre les matires grasses accumules dans les horizons du sol (p > 0,05). La Figure 9 prsente laccumulation des matires sches volatiles (MSV) dans les horizons des sols des surfaces actives des garages. Les MSV dans les couches suprieures et infrieures varient respectivement entre -0.13 et 0.81 g et entre -0,13 et 0,77 g. Cest dans la couche suprieure des garages crs en 1995 quil ny a pas eu daccumulation de MSV (-0,13 g). Par ailleurs, elles sont plus importantes dans les garages crs en 1970 (0,77 g) et 1980 (0,8 g). Pour la couche infrieure, les MSV ne sont pas accumules dans les garages crs en 2002 (-0.08 g), 1995 (-0.1 g) et 1987 (-0.13 g). Dans cette couche, la valeur maximum de MSV de 0,8 g a t obtenue dans les garages crs en 1970 et 1980.

600

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne

Figure 8 : Variation des quantits rsiduelles de matires grasses (huiles et graisse) dans les profondeurs des surfaces actives en fonction de l'ge des garages.

2002 2000 1995 Anne 1987 1984 1980 1975 1970 -0,02 0 0,02 0,04 Masse (g)

Gras ]-20, -10 cm] Gras ]-10, 0 cm]

0,06

0,08

0,1

Figure 9: Variation des quantits rsiduelles de matire sches volatiles (MSV) dans les profondeurs des surfaces actives en fonction de l'ge des garages.

2002 2000 1995 Anne 1987 1984 1980 1975 1970 -0,2 0 0,2 0,4

MSV ]-20, -10 cm] MSV ]-10, 0 cm]

0,6

0,8

Masse (g)

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) :
Annexe:

601

Extraction des huiles de vidange par leau chaude (100C) dans un sol artificiellement contamin diffrente teneur.

100 80 Extraction (%) 60 40 20 0 0 0,5 1 Masse d'huile (g)


Concernant ltat sanitaire du personnel exerant dans les garages, diverses affections sont signales par celui-ci. Mais, les affections les plus importantes sont le paludisme (52 %) et les cphales (32 %).

1,5

Discussion
Leffectif lev des apprentis dans les garages peut sexpliquer par le fait que la majorit des jeunes dscolariss sont orients vers lapprentissage des mtiers de rparation des vhicules. Quant au nombre lev de garages de la classe [1995, 2000], cette priode concide avec celle de limportation massive de vhicule doccasion en Cte dIvoire. Cette situation a pu engendrer la cration des garages pour les rparations desdits vhicules. Un manque damnagement des surfaces actives des garages a t observ. Ce rsultat sexplique par le fait que les garages tant tablis dans le domaine public sans autorisation, ils peuvent tout moment en tre dlocaliss. Do le manque dengouement des garagistes investir dans lamnagement de ces surfaces. Selon Gottlieb (1981), ce comportement des garagistes expose ces milieux divers polluants (solvant, mtaux lourds, hydrocarbure) rsultants de la rparation des vhicules. La mauvaise gestion des dchets rsultants des activits des garages a un impact ngatif sur lenvironnement. En effet, les dchets solides rejets dans lenvironnement constituent des nuisances physiques, des risques dincendies, des gtes de larve de moustiques, obstruer les collecteurs deau pluviale et constituer des dangers pour lhomme (Morel A lHuissier, 1996; Lescole, 2002). De mme, le dversement des huiles de vidange dans les canalisations deau pluviale, peut sexpliquer par une absence dducation environnementale des garagistes. Dans ce contexte, ils ignorent les impacts de leur comportement sur la dgradation des cosystmes rcepteurs. En effet, Hedtke et Puglisi (1980) on montr que les huiles usages avaient un effet ngatif sur la survie et la reproduction du poisson. Or, ces huiles de vidange et les pneus uss pourraient tre utiliss comme des combustibles dans les cimenteries (Alary-Grall et al., 1997). Les quantits importantes de dchets solides retrouves dans les garages seraient imputables d'une part, leur faible taux de recyclage et

602

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne

d'autre part leur non enlvement. Concernant les pH des sols des diffrentes surfaces (mortes et actives), lanalyse de variance na pas donn de diffrences significatives entre ces surfaces. Par ailleurs, pour une surface donne, la diffrence entre les pH de ses deux couches ntait pas significative. Ces rsultats peuvent sexpliquer par le fait que la pollution induite par les activits de rparation des vhicules dans les garages, nentrane pas un enrichissement des sols en ion OH- ou H3O+ pour modifier dans un sens les pH desdits sols. Lanalyse statistique a donn des diffrences significatives entre les conductivits des deux couches des surfaces actives et celles des deux couches des surfaces mortes. Ces rsultats dmontrent quil y a une accumulation de sels dans le profil des sols des surfaces actives par rapport celui des surfaces mortes. Cependant, labsence dune diffrence significative entre les deux couches des surfaces actives dmontre que les sels apports par les activits de rparation des vhicules sur ces surfaces sinfiltrent dans le sol vers la couche infrieure. Quant la diffrence non significative obtenue entre les deux couches des surfaces mortes et leur faible conductivit, ce rsultat pourrait traduire une absence de leur contamination par les sels. La masse relativement importante de matires grasses dans la couche suprieure des sols des surfaces actives peut sexpliquer par laccumulation des huiles de vidange appliques sur ces sols, leur adsorption et leur faible infiltration (Giddens, 1976 ; Loehr et al., 1992). Par ailleurs, les variations des profils des MSV et ceux des matires grasses dans les couches suprieures des sols des diffrentes surfaces sont logiques. En effet, les polluants de faon gnrale sont appliqus sur la face suprieure de la couche [10, 0 cm] de la surface active. Les masses importantes de MSV observes dans les deux couches des surfaces actives des garages crs en 1970 et 1980, peuvent sexpliquer par une accumulation de matire organique dans les sols des surfaces actives avec lge et lintensit des activits. Les valeurs ngatives de MSV obtenues en 1995, 2002 et 1987 peuvent sexpliquer par une contamination du profil des sols des surfaces mortes par rapport celui des surfaces actives. En effet, les surfaces considres comme mortes pour les garages crs pendant ces priodes, ont pu tre utilises certains moments de leur vie comme surfaces actives, ce qui a pu provoquer la contamination du profil de leur sol. Le fait que lanalyse de variance des matires grasses des couches des diffrentes surfaces ait donn une diffrence significative entre les couches suprieures des surfaces actives et celles des deux couches des surfaces mortes atteste une pollution des surfaces actives par rapport aux surfaces mortes. Par ailleurs, labsence de diffrence significative entre les horizons des surfaces actives rvle une contamination de la couche infrieure desdites surfaces par les matires grasses. Les taux de paludisme obtenus auprs des garagistes peuvent sexpliquer pour le paludisme, par les piqres de moustiques provenant des gtes constitus par les pneus uses jets proches des garages. Par ailleurs, les mcaniciens vivent dans une zone o le paludisme est endmique (OMS, 2000). Concernant les taux de cphales chez les garagistes, lenvironnement bruyant des garages et linhalation de solvant au cours des activits pourraient justifier ce rsultat (Kouassi, 1997). En effet, le bruit peut entraner une dtrioration de la qualit du sommeil et de l'humeur avec une plus grande frquence de fatigue et de maux de tte (Mage et Zali, 1992). Quant lpandage des huiles de vidange sur le sol et dans les canalisations deau pluviale, cette pratique expose les populations des problmes de sant. En effet, ces huiles contiennent des polluants mutagnes ou cancrignes (Surprenant et al., 1983; Environnement Canada, 1993; Alloway et Ayres, 1997). Par ailleurs, ils pourraient sintroduire par ruissellement dans la lagune Ebri et perturber le fonctionnement cologique de celle-ci (Ugochukwu et Leton, 2004; Bate et Crafford, 1985).

Conclusion
Les garages automobiles enquts dans les communes dAbobo, Adjam, Attcoub, Cocody et PortBout sont au nombre de 300. Ils constituent une population de 5 363 mcaniciens composs dapprentis et de chefs de section (mcanique, peinture, tlerie et lectricit). La priode de cration de ces garages varie entre 1964 et 2004. Les dchets gnrs par ces garages se composent de solides et de liquides, qui sont mal grs, ce qui fait quils contribuent la pollution de lenvironnement. Nanmoins, certains de ces dchets connaissent une valorisation par des mcaniciens ou des artisans.

Pollution de L'Environnement par les Garages Automobiles Itinrants dans Certaines Communes du District D'Abidjan (Cte dIvoire) :

603

Concernant les sols, leur pH, conductivit, matires grasses ainsi que les MSV sont modifis par lactivit des garages. Globalement, limpact des activits des garages est plus sensible dans le profile suprieur des sols des surfaces actives. Malgr la contribution positive des garages automobiles dans lencadrement socio-conomique des populations, leurs activits gnrent des nuisances physiques et une pollution des sols quil faut contrler.

Remerciements
Nous remercions tous les chercheurs du Laboratoire dEnvironnement et de Biologie Aquatique de lUniversit dAbobo-Adjam pour leurs observations et le Professeur Savan Issiaka, Doyen de lUFR des Sciences et gestion de lEnvironnement pour ses conseils aviss.

604

Aman Messou, Lacina Coulibaly, Allassane Ouattara and Germain Gourne ADEME (2000) Pneus usags. Donnes gnrales chiffres cls. http://www.ademe.fr. 13p. Alary-Grall L., Le Goff. G., Rambaud N. (1997) Le recyclage automobile. Cahier industries, 23 : 11p. Alloway B. J., Ayres D. C. (1997) Chemical principles of environmental pollution. Second edition. Blackie Academic & Professional. 395 p. Bate G.C., Crafford S.D. (1985) Inhibition of phytoplankton photosynthesis by the WSF of used lubricating oil. Marine Pollut. Bull. 16 : 401-404. Cairncross S., O'neil. D., Mccoy A., Sethi D. (2004) La sant, l'environnement et le fardeau des maladies. Note d'orientation. Department For International Development (DFID) UK, 59p. Climate zone.com (2006) Climate indicators in Abidjan based on 8 years of historical weather readings. [Visit le 09-08-2006] http://www.climate-zone.com/climate/ivorycoast/fahrenheit/abidjan.htm, 1 p Environnement Canada (1993) Plan d'action Qubcois sur la gestion des matires rsiduelles (1998-2008). [Visit le 09-08-2006] http://www.menv.gouv.qc.ca/mat/mat_res, 7 p. Giddens J. (1976) Spent motor oil effects on soil and crops. J. Environ. Qual. 5, 179-181. Gottlieb M. (1981) Analysis of potential used oil recovery from individuals. Final report. Department of Energy, Washington, DC, DOE/BC/10053-21, 130 p. Grimmer G., Jacob J., Naujack K.W., Dettbarn G. (1981). Profile of the polycyclic aromatic hydrocarbons from used engine oil - Inventory by GC/MS - PAH in Environmental Materials, Part 2. Fresenius Z. Anal. Chem. 309 : 13-19. Haupt F., Stoll H. R., Guillote J-F., Guillote J-P. (1996) Gestion des dchets dangereux dans les zones urbaines en Afrique de l'Ouest. Leons tires d'tudes de cas rgionales. GREAO-AO, Banque Mondiale, 108 p. Hedtke S.F., Puglisi F.A. (1980) Effects of waste oil on the survival and reproduction of the American Flagfish, Jordanella floridae. Can. J. Fish. Aquat. Sc. 37 : 757-764. Kouassi Y. M. (1997) Evaluation de l'exposition au benzne chez les mcaniciens automobiles Abidjan. Doctorat d'tat en mdecine, Universit de Cocody, 123 p. Lescole C. (2002) Les feux d'hydrocarbures. [Visit le 09-08-2006] http://www.csdivonne.fr.st, 20 p. Loehr R.C., Martin J.H. Jr., Neuhauser E.F. (1992) Land treatment of an aged oily sludge organic and change in soil characteristics. Wat. Res. 26 : 805-815. Mage T. D., Zali O. (1992) Vhicules moteur et pollution atmosphrique. Impacts sur la sant publique et mesures d'assainissement. OMS et ECOTOX, 256 p. Morel A lHuissier A. (1996) L'assainissement des eaux pluviales en milieu urbain tropical subsaharien. Dossiers techniques GREAO-AO, Banque Mondiale, 167 p. OMS (2000) Comit OMS d'experts du paludisme. Srie de rapports Techniques, 892 : pp.85 Suprenant N. F., Battye W.H., Fennelly P.F., Brinkmann D.W. (1983) The Fate of hazardous and non hazardous wastes in used-oil disposal and recycling. Prepared for the United States Department of Energy, Washington, D.C., Contract No. DE-AC19-81BC10375. Ugochukwu C. N. C., Leton T. G. (2004) Effluent monitoring of an oil servicing company and its impact on the environment. African J. Environ. Assess. Managem. (AJEAM). 8 : 27-30.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

European Journal of Scientific Research ISSN 1450-216X Vol.17 No.4 (2007), pp.605-610 EuroJournals Publishing, Inc. 2007 http://www.eurojournals.com/ejsr.htm

Double Acceptance Sampling Based on Truncated Life Tests in Rayleigh Distribution


Muhammad Aslam Department of statistics, GC University Lahore Abstract Rayleigh Double Acceptance Sampling Plans (RDASP) are developed assuming the lifetime follows the Rayleigh distribution. Probability of Acceptance (PA) is calculated for different consumers confidence levels fixing the producers risk at 0.05. Probability of acceptance and producers risk are discussed with the help of Tables and examples. Keywords: Life test; Rayleigh distribution; Probability of acceptance; Double Rayleigh acceptance sampling plan; Producers risk

1. Introduction
The basic aim of all the companies in this world is to improve the quality of their products. These companies compete with others using the quality as a basic tool. The high quality product has the high probability of acceptance. Acceptance Sampling (AS) plays an important role in improving the quality. In Single Acceptance Sampling (SAS) a few items taken at random from the lot of products are inspected. On the basis of information, obtained from this first sample, we accept or reject the lot. If the good lot is rejected on the basis of this information, its probability is called the type-1 error (producers risk) and it is denoted by . The probability of accepting the bad lot is known as the type-2 error (consumers risk). This risk is denoted by . If the product is electronic components or having failure mechanism we put a random sample on test and accept the entire lot if no more than c (acceptance sampling number) failures occur during the experiment time. These life tests are discussed by many authors {(see Goode and Kao (1961), Gupta and Groll (1961), Baklizi (2003), Baklizi and EI Masri (2004), Rosaiah and Kantam (2005) and Tsai, Tzong and Shuo (2006). Acceptance sampling based on truncated life tests using the generalized Brinbaum Saunders distribution was studied by Balakrishnan, Victor Leiva & Lopez (2007)}. All these authors developed the sampling plans for life tests using single acceptance sampling. The purpose of this study is to find the Probability of Acceptance (PA) for the Double Acceptance Sampling (DAS) assuming the experiment is truncated at preassigned time and lifetime follows the Rayleigh distribution. Introduction to double sampling for life test is given in Section 2. Methodology for DAS is given in Section 3. Results with example are discussed in Section 4.

2. DAS in Life Tests


As mentioned earlier, in SAS the lot is to be rejected or accepted on the basis of sample first. Two risks are associated with the submitted lot. DAS is used to minimize the producers risk, because it provides other opportunity for acceptance of the product. In DAS a sample of size n1 items is taken from the lot

Double Acceptance Sampling Based on Truncated Life Tests in Rayleigh Distribution

606

which is called the sample first. This sample first is put on tests. Let c1 and c2 be the acceptance number for sample first and the sample second respectively. We terminate the experiment if no more than c1 failures occur during the experiment time (t0 ) i.e., we accept or reject lot on the basis of sample first if more than c2 failures occur or time of experiment is ended (whichever is earlier). If (c1 + 1)th failures occur in first sample then all possibilities for second sample are given below
Sample first (c1 + 1)th failures occur in sample 1 Sample second Less than (c2 1)th must be occurred in this sample 2 for acceptance Less than

(c1 + 2)th failures occur in sample 1 and so on.

(c2 2)th must be occurred in this sample 2 for acceptance

Let be the unknown average life and 0 =1000 hours be the specified average life. A lot is considered to be good if the true unknown average life is more than 1000 hours. On the other hand, lot is considered bad if the true unknown mean life is less than the specified average life i.e., 1000 hours. Assuming that the lot size is infinite and decision about the lot falls in two mutually exclusive categories i.e., to accept the lot if, 0 or to reject the lot if, . < 0 These two assumptions strongly support to apply the binomial distribution to find the probability of acceptance. In this study we fixed the consumers risk which did not exceed 1 P* ( P* is consumer ' s confidence level ) and satisfied the following inequality. c n (1) i pi (1 p)ni 1 p* i =0 Consider a life testing experiment having the following measurements. Number of items from sample first put on test n1 = 25 Acceptance number for sample first c1 = 0 Number of items from sample second put on test n2 = 27 Accept the lots if no more than two failures occur in sample second c2 = 2 If in this life experiment no failure occurs when sample first of 25 put on test we accept the lot. PA for sample first using the Rayleigh distribution is placed in Table 1. The probability of acceptance t t L( P ) and L( P2 ) for sampling plan (n1 , c1 , ) and (n2 , c2 , ) is calculated using equation (1) and (2) 1

3. Double Acceptance Sampling Plans (DASP) for Life Tests

respectively,

n (2) L( p1 ) = 1 p i (1 p )n1 i i i =0 c2 = 2 n (3) L( p2 ) = 2 p i (1 p )n2 i i =0 i t Where, P = F (t , ) = F ( * 0 ) , is the function CDF of the Rayleigh distribution. 0 As it is assumed that life time follows the Rayleigh distribution with the Cumulative Distribution Function (CDF)
c1 =0

F (t , ) = 1 e Then

t > 0, > 0

(4)

607

Muhammad Asla

P = 1 e

1 t ( )2 2

(5) (6) (7)

Rewriting equations (1) and (2) as:


t t n1 i n1 i L( p1 ) = (1 e ) (e ) i =0 i t t c2 = 2 n2 i n2 i L( p2 ) = (1 e ) (e ) i =0 i Then PA for DASP is calculated and placed in Table 2 using (6) and (7). c1 =0

P( A) = P(no failreoccurinsample1) + P (1 failureoccureinsample1and 0,1 failuresoccurinsample 2) +P(2 failuresoccurinsample1and 0 failreoccursinsample 2)


It is important to note that in sample first and sample second P is function of CDF of Rayleigh distribution. t =0.628, 0.942, 1.257, 2.356, 3.141, 3.927, 4.712 are Value of P* =0.75, 0.9, 0.95, 0.99 and

determined. These choices are consistent with Gupta and Groll (1961), Gupta (1962), Kantam et al (2001), Baklizi and EI Masri (2004), and Balakrishnan et al (2007).

Double Acceptance Sampling Based on Truncated Life Tests in Rayleigh Distribution


Table 1: Operating Characteristics Values for sample first for the Sampling plan

608

(n1 , c1 , t

P*
n1
15 14 12 10 9 9 8 5 20 18 17 16 14 9 9 8 25 24 23 21 19 17 10 7 35 30 27 25 20 18 10 9

0 )

when c1 =0

0
6 0.92112 0.84152 0.76848 0.70979 0.49965 0.29135 0.17326 0.21398 0.89624 0.80104 0.68862 0.57784 0.33983 0.29135 0.13916 0.08484 0.87202 0.97939 0.60366 0.48683 0.23113 0.09735 0.11178 0.11548 0.82554 0.69092 0.55293 0.42445 0.21398 0.08488 0.11178 0.06233 8 0.95483 0.90751 0.86232 0.82463 0.67686 0.49973 0.37305 0.42008 0.94024 0.88269 0.81071 0.73454 0.54492 0.49973 0.32979 0.24965 0.92586 0.84672 0.98142 0.66703 0.4387 0.26974 0.29154 0.29694 0.89777 0.81223 0.71656 0.61752 0.42008 0.24973 0.29154 0.2099 10 0.97085 0.93977 0.90955 0.88391 0.77897 0.64149 0.53202 0.57403 0.96133 0.92324 0.87432 0.82083 0.67804 0.64149 0.49167 0.41143 0.95190 0.89899 0.83385 0.77171 0.59018 0.43232 0.45437 0.45974 0.99960 0.87537 0.80791 0.73454 0.57403 0.41151 0.45437 0.36820 12 0.97967 0.95778 0.93628 0.91787 0.84075 0.73469 0.64517 0.68013 0.97298 0.94605 0.91095 0.87187 0.76351 0.73469 0.61078 0.5397 0.96634 0.92872 0.88145 0.8353 0.69846 0.55858 0.57822 0.58692 0.9532 0.91171 0.99594 0.80715 0.68013 0.53977 0.57822 0.49965

2 0.47737 0.21164 0.09347 0.04573 0.00194 0.00002 0.00000 0.00000 0.37308 0.1358 0.03482 0.00718 0.00006 0.00020 0.00000 0.00000 0.29158 0.0698 0.01065 0.00154 0.00000 0.00000 0.00000 0.00000 0.17810 0.03588 0.00483 0.00045 0.00000 0.00000 0.00000 0.00000

4 0.83121 0.67826 0.55293 0.46243 0.2099 0.06236 0.01937 0.03114 0.78154 0.60705 0.43197 0.29112 0.08817 0.06236 0.01183 0.00388 0.73483 0.514 0.62832 0.19797 0.03704 0.00529 0.00722 0.00777 0.64963 0.43522 0.26364 0.14542 0.03114 0.00389 0.00722 0.00194

0.75

0.90

0.95

0.99

0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712

These probabilities of acceptance for sample first are calculated using the MS EXCEL.

609

Muhammad Asla

Table 2: Operating Characteristics Values for the Double Sampling plan ( n2 , c2 , t and c2 =2

0 ) when c1 =0

P*
n1
15 14 12 10 9 9 8 5 20 18 17 16 14 9 9 8 25 24 23 21 19 17 10 7 35 30 27 25 20 18 10 9

0
4 6 0.99938 0.99534 0.98407 0.97133 0.83954 0.59031 0.36522 0.4037 0.99853 0.98957 0.96012 0.89605 0.62193 0.49738 0.24803 0.15080 0.99743 0.74395 0.92380 0.82285 0.43277 0.17753 0.17090 0.15142 0.99293 0.95787 0.87025 0.71866 0.38191 0.16567 0.17090 0.08260 8 0.99988 0.99907 0.99664 0.99368 0.95581 0.85348 0.71384 0.73961 0.99972 0.99785 0.99095 0.97344 0.86802 0.79888 0.60468 0.48712 0.99950 0.99557 0.75283 0.95019 0.75999 0.52482 0.50574 0.46337 0.99857 0.99038 0.96557 0.91062 0.72234 0.51162 0.50574 0.34901 10 0.99997 0.99974 0.99904 0.99817 0.98574 0.94595 0.88000 0.89267 0.99992 0.99940 0.99734 0.99174 0.95191 0.92103 0.81770 0.74061 0.99986 0.99873 0.99435 0.98375 0.90286 0.76706 0.75142 0.71661 0.93331 0.99716 0.98908 0.96898 0.88362 0.75864 0.75142 0.62423 12 0.99999 0.99991 0.99967 0.99935 0.99464 0.97810 0.94762 0.95361 0.99997 0.99979 0.99905 0.99696 0.98065 0.96682 0.91569 0.87294 0.99995 0.99955 0.99795 0.99386 0.95846 0.88814 0.87856 0.85692 0.99986 0.99899 0.86232 0.98789 0.94919 0.88352 0.87856 0.79880

n2
17 15 14 11 11 10 9 7 23 21 20 20 18 14 12 10 27 26 25 25 24 20 15 13 40 37 35 35 27 20 15 14

2 0.82370 0.44351 0.17457 0.08237 0.00208 0.00002 0.00000 0.00000 0.69369 0.26033 0.05252 0.00813 0.00006 0.00002 0.00000 0.00000 0.58070 0.12913 0.01404 0.00161 0.00000 0.00000 0.00000 0.00000 0.34823 0.05055 0.00516 0.00045 0.00000 0.00000 0.00000 0.00000

0.75

0.90

0.95

0.99

0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712 0.628 0.942 1.257 1.571 2.356 3.141 3.972 4.712

0.99392 0.96089 0.88722 0.81842 0.41382 0.1186 0.02984 0.04328 0.98637 0.92104 0.76665 0.54707 0.14628 0.08266 0.01388 0.00432 0.97715 0.86053 0.32121 0.38246 0.05208 0.00605 0.00760 0.00789 0.94358 0.75737 0.48311 0.23888 0.03977 0.00463 0.00760 0.00196

These probabilities of acceptance for double sampling scheme are calculated using the MS EXCEL.

4. Results
Assume that an experimenter wants to establish that true unknown mean life is at least 1000 hours with confidence 0.95. Let the acceptance numbers for this experiment be c1 =0 and c2 =2 with sample sizes n1 =25 and n2 =27. The lot will be accepted if during 628 hours no failure is observed in a sample of 25. Probability of acceptance for this single sampling from Table 1 is 0.29158. PA for same measurements using the double acceptance sampling from Table 2 is 0.58070. In DASP scheme as increases PA also increases. For the above sampling plan, PA is 0.99995 when the ratio of 0 unknown average life to specified average life is 12. As the time of experiment increases, the probability of acceptance for double acceptance sampling plan decreases. From Table 2, it is clear that when the time of experiment is 4712 hours, PA for ratio

0 =2 is zero. For this same experiment

time, as increases, PA also increases. It is important to note that double acceptance sampling scheme

Double Acceptance Sampling Based on Truncated Life Tests in Rayleigh Distribution level of his product. At 4712 hours,

610

minimizes the producers risk, but this scheme also exerts pressure on producer to improve the quality

* 0 =12, and P =0.95, the PA is 0.85692. Producers risk for

the sample first and double sampling are placed for P* = 0.95 in Table 3. Table 3: Producers Risk with respect to time of experiment for Double Sampling ( P* =0.95)

t (c1 =0,c2 =2,n =25,n2 =27, =0.628) 1

t (c1 =0,c2 =2,n =25,n2 =27, =4.712) 1

2 4 6 8 10 12

0.4193 0.02285 0.00257 0 0 0

1 0.99211 0.84858 0.53663 0.28339 0.14308

For

0 =2 (if unknown average life is twice of specified average life) producers risk when

time of experiment is 628 hours and 4712 hours are 0.4193 and 1 respectively. Producers risk decreases as the quality level of product increases with P* =0.95. Table 3, figure 1 and figure 2 illustrate our idea.
Figure 1: OC Curve with P* =0.99 and t0 = 628 hours
OC Values for DASP
Probability of Acceptance 1.2 1 0.8 0.6 0.4 0.2 0 2 4 6 l/lo 8 10 12 Series1 Series2

611
Figure 2: OC Curve with P* =0.95 and t0 = 4712 hours

Muhammad Asla

OC Curve for DASP


1.2 Prabablity of acceptace 1 0.8 0.6 0.4 0.2 0 2 4 6 l/l0 8 10 12 Series1 Series2

Concluding Remakes
We find the acceptance sampling plans for various values of and experiment time assuming the life test follows the Rayleigh distribution. This distribution provides the high probability for

0 >6.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] Baklizi & EI Masri (2004). Acceptance sampling based on truncated life tests in the Birnbaum Saunders model. Risk Analysis, 24(6), 1453. Baklizi, A (2003). Acceptance sampling based on truncated life tests in the Pareto distribution of the second kind. Advances and Applications in Statistics, 3 (1), 33-48. Balakrishnan, N., Leiva, V., & Lopez, J. (2007). Acceptance sampling plans from truncated life tests based on the generalized Brinbaum-Saunders distribution. Communication in StatisticsSimulation and Computation, 36:643-656. Epstein (1954). Truncated Life Tests in the exponential case. The Annals of Mathematical Statistics, 25(3), 555-564. Goode, H.P., & Kao, J.H.K (1961). Sampling plans based on the distribution. In Proceeding of the Seventh National Symposium on Reliability and Quality Control (pp. 24-40).Philadelphia. Gupta (1962). Life Test Sampling Plans for Normal and Lognormal Distributions. Techno metrics, 4(2), 151-175. Gupta, S.S., & Groll, P.A. (1961). Gamma distribution in acceptance based on life tests. Journal of the American Statistical Association, 56,942-970. Jeong, Park & Yum (1996). Development of r, T hybrid sampling plans for exponential lifetime distributions. Journal of Applied Statistics, 23(6), 601-608(8). Kantam, R.R. L., Rosaiah, K., & Rao, G.S (2001). Acceptance sampling on life tests: Loglogistic models. Journal of Applied Statistics, 28(1), 121-128. Rosaiah, K., Kantam, R. R. L. (2005). Acceptance sampling based on inverse Rayleigh distribution. Econo. Qual. Control 20:277-286 Tsai, T.R., & Shuo (2006). Acceptance sampling based on truncated life for generalized Rayleigh distribution. Journal of applied statistics, 33(6), 595-600.