Beruflich Dokumente
Kultur Dokumente
&
Conference Dates
July 05-07, 2018
Conference Venue
Avrasya University, Trabzon, Turkey
ISBN
978-1-941968-47-5 ©2018 SDIWC
Published by
The Society of Digital Information and Wireless Communications (SDIWC)
Wilmington, New Castle, DE 19801, USA
www.sdiwc.net
Table of Contents
DIPECC2018
How Does Cloud Computing Change the Strategic Alignment Between Business and IT?……………… 1
TAEECE2018
Image Processing Based Anti-Sleep Alarm System for Drowsy Drivers ………………………………………….. 19
Peter Fuzes
Corvinus University Budapest
Budapest, Hungary
peter.fuzes@uni-corvinus.hu
strategy and goals of functional units, therefore offer limited customization options for their
improve the understanding of the business SaaS business applications [11, 15]. The public
domain. cloud SaaS model is based on provisioning
Impact of SaaS standardized solution to large number of
The key clients of SaaS services are the consumers with minimal customization.
business users. When purchasing an ERP, Limited room for customization does not allow
HCM (Human Capital Management), the business users to keep changing the
recruitment or marketing solution, business requirements; they have to accept the standard
users can articulate their needs and engage in solution.
direct discussion with the cloud supplier, who There is a trade-off between implementation
has experts with business knowledge on the time (and cost) and functionality. Accepting
specific solution field. Business users can standard solutions may lead to shorter
bypass company IT, and order business implementation cycle, however, the result may
application as a service (SaaS) from a cloud not be fully in-line with what the business users
provider directly. In this case, understanding want or need. The market success of SaaS
each other’s domain between the IT department solutions shows that large portion of customers
and business becomes irrelevant; the business; is willing to accept standardized solutions in
they can get access to business applications exchange for rapid and less expensive
faster directly from the SaaS provider. implementation.
Cloud providers can also influence the
4.3 The Expressed Need of Business customer needs by sharing best practices and
Keeps Changing experiences with other customers. This can
Business needs may change during an IT help the business users to realize their latent
project due to the change in the competitive needs earlier during the implementation process
environment [4]. Also, business users may or give up requirements for unnecessary
realize during the implementation project that features.
the IT solution can deliver additional value for Overall, using cloud-based business
them with extensions or customization, and as a applications (SaaS) can have a positive impact
result they may change the requirements. In on preempting and limiting the change of
such situation, IT is ‘shooting to a moving expressed business needs.
target’.
Even if the necessary budget is available to 4.4 IT Systems Are Not Flexible Enough
address the changed business need, the on- to Support the Changing Need of
premise IT system may not be flexible enough Business
to accommodate the required changes short The limited capability of the on-prem IT
term, which can negatively impact the infrastructure a key reason for the inflexibility
business-IT alignment. of company IT systems. The cycle of allocating
Impact of IaaS and PaaS financial resources, run the procurement
Using IaaS and PaaS cloud service makes the process and implement the new hardware or
IT system more elastic and allows to rapidly software parts may take a long time. That long
scale up and down. With IaaS and PaaS, the IT cycle can be seen by business users as
department can respond to the changing need of inflexibility and lack of ability to address the
the business better than with on-prem business needs.
environment, therefore the alignment can Impact of IaaS and PaaS
improve. Cloud providers build large capacity data
Impact of SaaS centers, and from those data centers they can
Increased elasticity also applies to SaaS rapidly scale up and down the service allocated
solutions. However, SaaS has another aspect to the customers. Therefore, when the company
which can help to limit the customization IT department uses IaaS and PaaS services, the
required by the business users. Cloud providers existing IT infrastructure capacity will not be a
limiting factor to serve the business user’s IT department’s loss of control over the
needs. Cloud provides elasticity, which is not company’s IT system can increase the ‘chaos’
the case with on-prem solutions [16]. As a of existing systems, instead of decreasing it.
result, IT systems become more flexible from Without coordination by the IT department, the
the business point of view and able to support Enterprise Architecture of the company may
the changing business needs, thus the alignment disintegrate [18, 19], which may have a
may improve. negative impact on the company’s operation.
Impact of SaaS Therefore, it is imperative for the IT
Another reason for IT inflexibility is the department to become a service integrator in a
complexity and inhomogeneity of existing new governance model, linking and integrating
systems. When integration is required with the different on-prem and cloud services.
existing systems, cloud services do not resolve The appearance of ‘shadow IT’ - when business
this problem, however, those can be used for a users directly ordering cloud services from
new application outside of the current cloud providers without involving the IT
architecture. department – is not only a possibility, but
For example, a SaaS solution for recruiting new already reported on the market. Based on the
employees (advertising available positions on- response from 245 companies, a survey run by
line, accepting and sorting uploaded CVs, Select Hub concludes that “There is a
providing a workflow for the selection process) movement away from allowing the experts in
does not necessarily have to be closely IT departments to exercise complete control
integrated with the company’s ERP system. (due to their expertise), and turn towards the
Instead of waiting for the upgrade of the on- users themselves, as the software environment
premise ERP system to add a new recruitment encourages their participation” [20].
module to it, the HR department may use a Tim Killenberg, senior vice president of N3 (an
recruitment SaaS solution from an external outsourced, integrated sales and marketing
provider, without dealing with the problem of execution firm), says that “We are seeing a
inhomogeneity of the internal IT systems. sharp increase in the number of line-of-business
With SaaS, new business solutions can be buyers who are empowered to make technology
provided in a shorter timeframe, which purchase decisions themselves. Reaching these
improves the alignment between the business buyers requires a new sales mindset and new
needs and the IT solutions supporting those skills” [21].
needs.
6 CONCLUSION
Based on the analysis, Table 1 summarizes the The role of the company’s IT department is
possible impact of cloud services on strategic changing due to the growing use of cloud
alignment problems. computing services, and this change could
positively impact some of the problematic areas
5 DISCUSSION
of business-IT alignment.
Based on our analysis, the use of cloud services IT department has an important role to
may improve the alignment between business coordinate and integrate between different on-
and IT. However, changing role of the IT premise and cloud-based solutions. By
department poses a threat as well: coordination outsourcing basic tasks to cloud providers and
of different IT systems may fall out of their reassigning roles, IT department has better
hand. There is a risk that several systems will possibility to focus on higher level business
be used by the company – as result of the direct solutions and to become the strategic partner of
purchase by business units without the the business.
involvement of IT department – which will not
be integrated or linked. The appearance of
shadow IT can increase the risk of loss of
valuable and confidential data [17].
Mokhled S. AlTarawneh
Mutah University
Faculty of Engineering, Computer Engineering Department P.O.Box (7),
Mutah 61710, Jordan
mokhled@mutah.edu.jo
University (HU), AlBalqa Applied University these will include, but not be restricted to:
(BAU)); Southern region (Mutah University [12] claimed that E-learning system based on
(MU), Tafelah technical university (TTU), CC infrastructure is feasible and it can greatly
Al-Hussein Bin Talal University (AHU)). improve the efficiency of investment and the
This paper is divided into five main sections. power of management, which can make E-
This section introduces the main theme of the learning system development into a virtuous
paper. A previous work is given in Section 2. circle and achieve a win-win situation for
Section 3 presents a detailed description of suppliers and customers. [13] show that, it is
proposed Architecture E-learning center of essential for an educational and learning
excellence based on private Jordanian organization, with its budget restrictions and
Universities. Finally, in Section 5, sustainability challenges, to use the cloud
conclusions recommendations for future work formations best suited to its IT activities. A
are pointed-out. cloud computing formation model was
proposed. [14] focused on the ways of
2 PREVIOUS WORKS applying cloud computing to transfer
traditional e-Learning to a cloud based e-
Cloud computing is being deployed in many Learning with overall half less costs (i.e.
organizations as well as in educational implementation, maintenance, update and re-
institutions, research in this field could be engineering). [15] described the concept of e-
categorized into two groups: education as learning and cloud computing, how to
service and learning as service. In the first implement it, the efficient utilization of
group, [8] survey the state of the art on the hardware resources and software in a learning
use and research of cloud computing in environment.
education taking in account the main
educational stakeholders, exemplified and 3 PROPOSED ARCHITECTURE
categorized the advantages and risk of CC in
education, identify and discuss promising Private cloud is proposed to use benefit of
technical issues following with systematic existed, developed and maintained services
methodology, they reviewed 112 scientific and infrastructure on a Jordanian private
works on cloud computing in education up to network, it could be built on Jordanian
year 2012. [9] investigated whether there is a Universities Network (JUNET) infrastructure;
need between students for applications and therefore, the adoption of E-Learning center
services in the “cloud” software as services of excellence (ELCE) could be established on
(SaaS), the extent to which they use them and geographical determined regional covered
what types of applications and services are area, i.e. to chose one data center in north part
leading. They analyzed and interpreted the of Jordan as well one in middle another in
results which provide indications of students' southern region, and according to technical
willingness to “move to the cloud”. [10] readiness and geographical location, JUST,
discussed the importance of online training JU and MU could be suggested for north,
and emphasizes on its qualitative and central and southern region respectively,
quantitative development for some where the north part of Jordan covered Irbid,
organizations. They concentrated on utilizing Ajloun, Jerash and Mafraq governorates,
online education based on cloud computing central part covered Balqa, Amman, Zarqa
environments. They discussed the necessity of and Madaba, while southern part covered
cloud-based educational systems for Karak, Tafilah , Maan and Aqaba, Figure 1.
organizations and countries. [11] reviewed the Using this scenario will raise the
adoption of CC in education in developing availabilities, reliabilities of planning
countries and Arabic countries in particular. services. This scenario will be used as a
The second group of literature survey, e- redundant infrastructure for all universities
learning as service tackled in many articles, members; as well they will use benefit of
replication approaches to reduce the service increasing the connection bandwidth and
risk. developing their data centers to achieve
scalability, persistent storage, distributed
access, efficient resource usage and
interoperability of the E-learning center of
excellence objects. The proposed private JU
cloud offers greatest level of privacy,
security, reliability and control, but it is
required the core data centers to purchase and
maintain all needed software, licenses‟ and
needed infrastructure hardware which slightly
increases the initial cost of proposed cloud.
Abstract—In this paper, a back-to-back (BTB) PSC-PWM technique [11]. Combining the averaging
modular multi-level converter (MMC) is designed for control with capacitor voltage balancing control
a permanent magnet synchronous generator (PMSG) enables capacitor voltage balancing issue without any
based wind energy conversion system (WECS). The external balancing circuit.
switching of power semiconductors of BTB-MMC is
based on phase-shifted carrier pulse width modulation II. PMSG BASED WECS DESIGN
(PSC-PWM). Capacitor voltage balancing issue of the
Two MMCs can be connected from their DC link
BTB-MMC is realized with a control system working
to form a BTB arrangement. Thus, BTB-MMC can be
in accordance with PSC-PWM technique. The
used in a WECS. The MMC near wind generator
designed PMSG based WECS is verified through case
operates as an AC-DC converter, while the other one
studies using PSCAD/EMTDC.
is connected to AC grid and operates as a DC-AC
converter. By this way, the real power generated by
Keywords: Permanent magnet synchronous
the wind turbine is transferred to the AC grid in a
generator; Wind energy conversion system; Modular
controllable way. The sub-module (SM) or cell is the
multi-level converter; Back-to-back converter.
fundamental building block of an MMC. Each SM can
be constructed by a half-bridge (HB) having two
I. INTRODUCTION power semiconductors (IGBT) and a capacitor [12],
[13]. Fig. 1 shows the switching states of a SM. Table
With the quick exhaustion of conventional fossil
1 lists the SM capacitor cases. The PMSG based
fuels, many renewable energy sources have emerged
WECS design using a BTB-MMC arrangement is
as substitutions for conventional sources [1]. In recent
shown in Fig. 2. Each MMC has three legs and each
years, the wind energy has become the fastest
leg contains an upper and a lower arm [12], [14]. Four
expanding renewable source [2], [3]. The permanent
sub-modules (SM) or cells are connected in series in
magnet synchronous generator (PMSG) is one of the
each arm. With this design, line-to-line AC voltage of
growing trend in wind energy conversion system
each MMC has 2n+1=17 levels, where n is the
(WECS) design because of many advantages such as
number of cells per arm).
no need for gearbox and rotor excitation, brushless
structure, low maintenance cost/weight, and wide
range of speed control [4],[5],[6]. It is necessary to use S1 D1 S1 D1
+ C Vcmp + Vcmp
Vjp S2 D2
transformer [9]. The MMC was first proposed by (c) Case 3 (i<0) (d) Case 4 (i<0)
Marquardt, R. for high voltage direct current Fig. 1. Switching states of SM.
transmission [7]. MMC is very suitable for medium
voltage/high power WECS applications due to modular Table 1. SM capacitor cases.
circuit topology and independent control capability of Current direction SM state Capacitor case
each MMC [10], [11]. In this paper, a back-to-back Positive i>0 ON Charge
(BTB)-MMC topology is applied to a PMSG based Negative i<0 OFF Bypass
WECS. On the basis of the works realized in [4], the Positive i>0 OFF Bypass
control strategy of the BTB-MMC is realized based on Negative i<0 ON Discharge
Phase-shifted carrier pulse width modulation As shown in Fig. 2, 𝑖𝑝𝑣 is a positive current of phase-
(PSC-PWM) has been considered in this work due to v, and 𝑖𝑛𝑣 is a negative current of phase-v. From the
easy capacitor voltage balancing, equally distributed block of the averaging control in Fig. 3, the circulating
switching frequency in each SM [16], low harmonic current 𝑖𝑧𝑣 , can be defined as:
distortion, and lower semiconductor losses [15], [16].
𝑖𝑣 𝑖𝑣
PSC-PWM has two waveforms. If the triangular 𝑖𝑧𝑣 = 𝑖𝑝𝑣 - = 𝑖𝑛𝑣 +
2 2
carrier is higher than the reference waveform, the SM 1
is bypassed, otherwise, the SM is inserted. The phase = (𝑖𝑝𝑣 + 𝑖𝑛𝑣 ) (2)
2
shift between these two waveforms is 2π/n. PSC-
PWM technique needs to accomplish two tasks 𝑖𝑧𝑣 is a loop current that is impossible to measure
(average control and balancing control) in order to directly whereas 𝑖𝑣 is a branch current [11]. As shown
complete its function completely. in Fig. 3, the 𝑖𝑧𝑣𝑔 is the current command of the 𝑖𝑧𝑣 ,
and can be defined by
A. Average Control
The circulating current occurs due to the 𝑖𝑧𝑣𝑔 = 𝑘1 ( 𝑣∗ 𝑐 - 𝑣𝑐𝑣𝑔 ) + 𝑘2 ∫( 𝑣 ∗ 𝑐 − 𝑣𝑐𝑣𝑔 )𝑑𝑡 (3)
imbalance of the average capacitor voltages between
the phases of the MMC [9], [11]. The circulating The driving voltage 𝑣𝐴𝑣𝑔 acquired from the averaging
current has no influence on the AC or DC side; it only control is shown in Fig. 3, given by
flows through the phases of the MMC. Consequently,
the circulating current causes power losses and ripples 𝑣𝐴𝑣𝑔 = 𝑘3 (𝑖𝑧𝑣 - 𝑖𝑧𝑣𝑔 ) + 𝑘4 ∫(𝑖𝑧𝑣 − 𝑖𝑧𝑣𝑔 )𝑑𝑡. (4)
observed in capacitor voltages of the SMs [15], [17].
The arm inductors as shown in Fig. 2 are designed to When 𝑣𝑐𝑣𝑔 ≤ 𝑣∗ 𝑐 , 𝑖𝑧𝑣𝑔 increases. The current miner
lower the circulating current. Fig. 3 shows the overall
control scheme of each MMC, including DC capacitor loop is used to force the actual DC-loop current 𝑖𝑧𝑣 to
voltage control and circulating current control for v- follow the 𝑖𝑧𝑣𝑔 .
phase in AC-DC MMC. The control scheme for other
phases is the same. Fig. 3 shows the block of the B. Balancing Control
averaging control. It forces the v-phase average
voltage 𝑣𝑐𝑣𝑔 to follow its command 𝑣∗ 𝑐 , where the Fig. 3 shows the balancing control block for
phase-v. The balancing control is used to force the
𝑣𝑐𝑣𝑔 is given by
individual DC voltage to follow its command 𝑣∗ 𝑐 [11].
As shown in Fig. 3, the gained output voltage from the
balancing control is 𝑣𝐵𝑚𝑣𝑔 . The balancing control is As shown in Fig. 3, 𝑣∗ 𝑚𝑣 is the AC-voltage that is
based on the phase current that is either positive or gained from the balancing after passing through
negative. The polarity of 𝑣𝐵𝑚𝑣𝑔 must be due to 𝑖𝑝𝑣 or positive and negative command arm. Every SM
𝑖𝑛𝑣 [17], [18]. When 𝑣∗ 𝑐 is greater than or equal the capacitor voltage 𝑣𝑐𝑣𝑚 normalizes voltage
𝑣𝑐𝑣𝑚 (m: 1−4) in the upper arm. The positive active command 𝑣 ∗𝐴𝑣 [11], [17].
power has to be taken from the DC-link into the four
IV. SIMULATION STUDIES
SM. When 𝑖𝑝𝑣 is positive, the product of 𝑣𝐵𝑚𝑣 and 𝑖𝑝𝑣
generates the positive active power. When 𝑖𝑝𝑣 is BTB-MMC based PMSG WECS, presented in
negative, the polarity of 𝑣𝐵𝑚𝑣𝑔 must get reverse to Fig. 1, is simulated in PSCAD/EMTDC environment.
occupy the positive active power. At last, for m =1~4 The model parameters used for simulation are listed in
is represented as: Table 2. The data of the wind turbine to produce the
mechanical torque to the PMSG are given in Table 3.
𝐾5 ( 𝑣∗ 𝑐 − 𝑣𝑐𝑣𝑚 ) (𝑖𝑝𝑣 > 0) The output of the WECS is connected to the AC grid
𝑣𝐵𝑚𝑣𝑔 = { ∗ (5) whose parameters are given in Table 4.
−𝐾5( 𝑣 𝑐 − 𝑣𝑐𝑣𝑚 ) (𝑖𝑝𝑣 < 0)
Table 2. Model parameters used for simulation.
where m, is the number of cell. However the equation Rated active power 5 MW
of 𝑣𝐵𝑚𝑣𝑔 for m = 5 ~ 8 is: Rated frequency 50 Hz
DC-link voltage 9 kV
𝐾5 ( 𝑣∗ 𝑐 − 𝑣𝑐𝑣𝑚 ) (𝑖𝑛𝑣 > 0) arm inductance 3 mH (6.4%)
𝑣𝐵𝑚𝑣𝑔 = { (6)
SM capacitor voltage 2.25 kV
−𝐾5( 𝑣∗ 𝑐 − 𝑣𝑐𝑣𝑚 ) (𝑖𝑛𝑣 < 0)
Carrier frequency 2 kHz
The voltage command of every sub-module 𝑣∗ 𝑚𝑣 is Coupling inductance 6 mH
shown below for positive-arm and negative-arm
Table 3. Wind turbine parameters.
commands:
Wind speed 12 m/m
𝑣∗ 𝐸 Frequency 50 Hz
𝑣∗ 𝑚𝑣 = 𝑣∗ 𝐴𝑣 + 𝑣𝐵𝑣𝑚𝑔 - 𝑣
+ (m: 1 ~ 4) (7) Rotor radius 35 m
4 8
𝑣∗ 𝑣 𝐸
𝑣∗ 𝑚𝑣 = 𝑣∗ 𝐴𝑣 + 𝑣𝐵𝑣𝑚𝑔 + + (m: 5 ~ 8) (8) Active power 5 Mw
4 8
Table 4. AC grid parameters. Fig. 5 (a) Active and reactive power output of PMSG, (b) DC-link
Frequency 50 Hz voltage of BTB-MMC, (c) Circulating current.
Rated voltage L-L rms 5.5 kV Fig. 5 shows the active and reactive power output of
the PMSG, DC-link voltage of BTB-MMC, and
In this work, the controllers of the WECS are circulating current, respectively. As shown, the
examined under two cases: under steady-state rated circulating currents are minimized. Fig. 6 shows the
conditions and under dynamic state when DC-link active and reactive power of DC-AC MMC and line-
voltage is dropped. to-line voltage of DC-AC MMC, respectively. From
these waveforms, we can see that the DC-AC MMC
A. Case 1: Under Rated Conditions side is able to output 17 levels at the line-to-line
voltage converted from the DC-link voltage that has
Fig. 4 shows the simulated waveforms of line-to- been delivered by AC-DC MMC.
line voltages, phase currents, and line-to-neutral
voltages of AC-DC MMC under rated conditions,
respectively.
(a)
(a)
(b)
Fig. 6 (a) The active and reactive power of DC-AC MMC, (b) Line-
to-line voltage of DC-AC MMC.
(b)
In order to generate high quality AC voltage, the DC
capacitor voltage of SM per each arm must be equal.
Fig. 7(a) shows the DC capacitor voltages of eight
SMs in MMC1. As observed, DC capacitor voltages in
phase-u of MCC1 has no ripples. Fig. 7(b) shows the
DC capacitor voltages of eight SMs in phase-w of
(c)
Fig. 4 (a) Line-to-line voltage of AC-DC MMC, (b) Phase currents MMC2. Theoretically, DC capacitor voltage of each
of AC-DC MMC, (c) Line-to-neutral voltage of AC-DC MMC. SM is 2.25 kV. This value can be calculated as:
DC supply voltage 9
𝑣∗ 𝑐 = = (9)
number of SM in upper arem or lower 4
(a)
(a)
(b)
(b)
Fig. 7 (a) DC capacitor voltages of AC-DC MMC, (b) DC capacitor
(c) voltages DC-AC MMC.
B. Case 2: Under Dynamic Case has returned to their nominal values after some
disturbances.
In this case study, the dynamic performance of the
control scheme is examined when DC link voltage set
point is dynamically changed for a duration of 0.05s
as shown in Fig. 8(a). As observed from the simulated
waveforms presented in Fig. 8, the control system of
the WECS was able to ride through these dynamic
states smoothly.
(a)
(a)
(b)
(b)
(c)
Fig. 9 (a) DC-capacitor voltages of MMC1, (b) DC-capacitor
voltage of MMC2, and (c) voltages of AC three phase voltage source
V. CONCLUSION
The design and control of the power converters in
(c) a WECS is still challenging. The design and
simulation of BTB-MMC based PMSG WECS is
proposed in this work. BTB-MMC arrangement brings
many advantages to WECS design such as being
flexibility and modularity. Moreover, MMC is able to
generate higher voltages with diminished voltage stress
on each power electronic semiconductor. Harmonic
(d) levels can also be significantly decreased thanks to the
increased number of voltage levels. As a result of this,
filtering requirements are becoming less. MMCs have
also better fault ride through capabilities. PSC-PWM
method is used to trigger power semiconductors in
each SM. Balancing and average control methods are
designed to achieve the balancing of capacitor voltage
(e)
Fig. 8 (a) DC link voltage set-point change, (b) Line to line voltage in each SM in both MMCs. In order to verify the
of MMC1, (c) Active and reactive power of DC/AC side, (d) Phase efficiency of the proposed control scheme, two cases
currents of MMC2, (e) Active and reactive power of PMSG. studies were performed in this work. As observed from
the simulation results, the dynamic performance of the
After the drop of DC link voltage, three-phase AC proposed control scheme is satisfactory.
voltage of both MMCs are also diminished. Since, the
DC link voltage is used as reference voltage in the REFERENCES
controller for both MMCs. Thus, the supplied and
absorbed active power of MMC1 and MMC2 are [1] Yaramasu, Wu, Sen, Kouro, and Narimani, "High power Wind
Energy Conversion Systems: State-of-the-Art and Emerging
respectively decreased. Fig. 9(a) shows the voltages of Technologies, " IEEE. Vol. 103, No. 5, May 2015, pp. 736-
eight SMs in phase-w of MMC1 when DC link voltage 739
set-point change is applied. The voltages of each SM
[2] Henk, P., Jan Abraham, "Trends in Wind Turbine Generator [17] Suman Debnath, Jiangchao Qin, Behrooz Bahrani, Maryam
Systems,” IEEE Jornal of emerging and selected topics in Saeedifard, and Peter Barbosa, “Operation, Control, and
power electronmics Vol. 1, No. 3, Sep 2013, pp. 174-185 Applications of the Modular Multilevel Converter: A
[3] Yaramasu, Wu, Sen, Kouro, and Narimani, "High-Power Review,” IEEE Transactions on Power Electronics., vol. 30,
Wind Energy Conversion Systems: State-of-the-Art and Jan. 2015, pp. 37–53
Emerging Technologies,” Proceedings of the IEEE., Vol. 103, [18] Hirofumi Akagi, Shigenori Inoue, and Tsurugi Yoshii,
No. 5, May 2015, pp. 736-739 “Control and Performance of a Transformerless Cascade
[4] Mian, W., Yaowei, H., Wenjian, Z., Yue W, Guozhu C, PWM STATCOM With Star Configuration,” IEEE
"Application of modular multi level converter in meduim Transactions on industry Application, VOL. 43, NO. 4,
voltage high power permanent magnet synchronous generator July/August 2007, pp. 1041-1049
wind energy conversion system,” IET Power. GEN.
Distrib.,Mar 2016, pp. 1–10
[5] M.S. Camara, M.B. Camara, B. Dakyo and H. Gualous,
"Permanent Magnet Synchronous Generators For Offshore
Wind Energy System Linked To Grid Modeling And Control
Strategies", IEEE, 16th International Power Electronics and
Motion Control Conference, PEMC2014, Antalya/Turkey,
September, 2014
[6] Tan Luong, Thanh Dung, Thanh Trang, and Huy Dung
"Advanced Control Strategy of Back-to-Back PWM
Converters in PMSG Wind Power System,” Power
Engineering and Electrical Engineering. Vol. 13, June, 2015,
pp. 81-95
[7] A. Lesnicar and R. Marquardt, "An innovative modular
multilevel converter topology suitable for a wide power
range," in Proc. 2003 Power Tech Conference Proceedings,
2003 IEEE Bologna
[8] W. Abd Halim, S. Ganeson, M. Azri, T.N.A. Tengku Azam,
“Review of Multilevel Inverter Topologies and Its
Applications,” Journal of Telecommunication, Electronic and
Computer Engineering, Vol. 8 No. 7, pp. 51-56
[9] Makoto Hagiwara, Kazutoshi Nishimura, and Hirofumi
Akagi, “A Medium-Voltage Motor Drive With a Modular
Multilevel PWM Inverter,” IEEE Trans. Power Electronic.,
Vol. 25, No. 7, Jul 2010, pp. 1786-1799
[10] Ahmed, N.; Haider, A.; van Hertem, D.; Zhang, L.; Nee, H.-
P. Prospects and challenges of future HVDC Super Grids with
modular multilevel converters. In Proceeding s of the 14th
Conferenceon Power Electronics and Applications (EPE
2011), Birmingham, UK, 30 August–1 September 2011; pp.
1–10.
[11] M. Hagiwara and H. Akagi, “Control and experiment of pulse
width modulated modula rmultilevel converters,” IEEE Trans.
Power Electron., vol. 24, no. 7, Jul. 2009, pp. 1737–1746
[12] H. Vahedi and K. Al-Haddad, "Half-bridge based multilevel
inverter generating higher voltage and power," in Electrical
Power & Energy Conference (EPEC), 2013 IEEE, 2013, pp.
1-6. Makoto Hagiwara, and Hirofumi Akagi, “Control and
Experiment of Pulsewidth-Modulated Modular Multilevel
Converters,” IEEE Trans. Power Electron., Vol. 24, No. 7, Jul
2009, pp. 1737–1746
[13] R. H. Baker, "Switching circuit," ed: US Patent 4,210,826,
1980
[14] Pracha Khamphakdi, Kei Sekiguchi, Makoto Hagiwara, , and
Hirofumi Akagi, “Design and Experiment of a Back-To-Back
(BTB) System Using Modular Multilevel Cascade Converters
for Power Distribution Systems” IEEE Trans. Power Electron,
2013, pp. 311-317
[15] Debnath, S., Qin, J., Bahrani, B., et al.:‘Operation, control, and
applications of the modular multilevel converter: a review’,
IEEE Trans. Power Electron., 2015, 30, (1SI), pp. 37–53
[16] Manish V. Kurwale ,Palak G. Sharma , Gautam Bacher,
“Performance Analysis of Modular Multilevel Converter
(MMC) with Continuous and Discontinuous Pulse Width
Modulation (PWM) ,” International Journal of Advanced
Research in Electrical, Electronics and Instrumentation
Engineering., Vol. 3, Issue 2, February 2014, pp. 7463-7474
Abstract—In this paper, we propose an underwater acoustic shift results loss of information and frame desynchronization,
communication based on trigonometric chirp modulated as well. In addition to relative speed between the source and
waveforms. We show that a usual system with linear chirp the receiver, different sampling frequencies also yield to signal
modulation requires wide bandwidth. Therefore, a Doppler
resilient and robust underwater digital communication system compression-expansion. This means that for example, if our
based on a non-linear frequency modulation as trigonometric DAC is not accurate, we have to consider this problem in the
chirp technique has been presented. system.
To evaluate the performance of trigonometric chirp modulation Recently, there have been considerable attempts to design
technique over AWGN channel, Rayleigh fading and Rician robust receivers which prosperously equalize the channel and
channels with and without Doppler effect, we have conducted
some simulations. It is demonstrated that the performance of also increase transmission rate.
proposed method, even in lower spreading factors gives better Chirp spread spectrum (CSS) techniques use chirp signals
performance than an LFM chirp modulated system in terms of for data transmission. Chirp is a sinusoidal signal whose
bit error rate. frequency decreases or increases over a specified time
duration. Chirps as another spread spectrum signals are
Keywords—Underwater Acoustic Communication, Linear commonly used in radar applications and many years later in
frequency Modulation (LFM), Trigonometric Chirp, Doppler sonar due to its good temporal resolution of auto correlation
Shift, Rayleigh Fading, Rician Fading.
function and also, considerable processing gain obtained
from correlation measurement in the receiver [1]–[9]. In
I. I NTRODUCTION
march 2007, IEEE introduced CSS physical layer in its new
Over past decades, many underwater wireless standard as 802.15.4 a, allowing CSS to be used in various
communication technologies have been proposed and application such as industrial control, sensor networking, real
applied as ocean exploration, oceanography data collection, time location systems and medical devices [10].
control over autonomous underwater vehicles, undersea Some characteristics of chirp modulations make it robust to
navigation and etc. multipath with low Doppler sensitivity. We know that linear
When studying problems of sending data from transmitter frequency modulation (LFM) is the simplest chirp waveform.
to receiver, the distortion induced by the channel lead us It has some advantages in comparison to non-linear chirps such
to analyze for then being able to choose a right signal as easy generation by different technologies, mostly simple to
waveform. A reliable underwater acoustic communication is process by a matched filter or similar techniques. Accordingly,
challenging because of some channel characteristics such as; more applications for the CSS system based on linear chirps
small bandwidth, high power attenuation, fast time variation have been developed than that for non-linear types. But the
of the channel response, fading along of multipath propagation main drawback is obtaining orthogonality between two LFM
and Doppler effect due to relative speed between the receiver signals. A large time-bandwidth product, known as spreading
and the transmitter. factor, is needed. Either considerable symbol duration or wide
Actually in UWA communication, due to the low velocity bandwidth causes large spreading factor. The time-bandwidth
of acoustic waves (roughly 1500 m/s) Doppler shift is larger product of a linear CSS should be more than 70 to achieve
than terrestrial radio frequency communication. But this is relatively acceptable orthogonality between its symbols for
not the only impact of Doppler in acoustic waves. Unlike binary signaling. As we know, significant bit rate is more
the case of radio transmission which Doppler shift is usually preferable [11]. Thus, by necessity a classical LFM signal will
modeled as a frequency offset, the effect of Doppler shift on occupy undesirable large bandwidth especially at high speed
the symbol duration cannot be neglected. Significant Doppler data rate. So it is important to choose a pair of chirp signals
which will be sufficiently orthogonal in small spreading factors fc is the carrier frequency. Both signals have a common
to sending information. Nevertheless, non-linear chirps have bandwidth B, and Ts is the chirp duration in second. We
not attained high prevalent like linear type, so far. Probably it will send s0 as a bit 0 and s1 as a bit 1. At the receiver
is because of limited availability of non-linear-FM generation as shown on Fig.1, where an AWGN channel considered with
and processing devices, more complexity of system, derivation n ∼ N (0, N20 ), two decision variables are produced:
of performance evaluation can be troublous, as well. By
Γi = khr, si ik, i = 0, 1 (2)
accepting these issues, desired performance can be achieved
with considerable reduction in bandwidth. Better performances Where kh∗, ∗ik is the absolute value of the inner product
due to some intrinsic properties of non-linear chirps were between observation r and transmitted signal si . So, we can
motivation of our research. consider the hypothesis test as follows
In this paper, we use trigonometric chirp signals to modulate
binary data. To obtain a better performance, it is preferable H0 : r = s0 + n
(3)
to consider two orthogonal chirp signals. At the receiver, H1 : r = s1 + n
we consider a matched filter to correlate the received signal
According to the non-coherent maximum a posteriori
with a known signal. We illustrate that a trigonometric chirp
(MAP) detector, the decision rule will be
modulated system has better performance over AWGN and
H0
different fading channels in contrast with a system based on khr, s0 ik − khr, s1 ik>
<H1 0 (4)
LFM chirp.
The remainder of this paper is organized as follows. In section Choosing H1 when H0 occurs and also vice versa, means
II, the orthogonality of LFM and trigonometric chirp as a type an error happens. By totally analyzing expression (4), we
of non-linear frequency modulation have been investigated. In know that inner product of LFM signals is not equal to zero
section III, performances over additive white gaussian noise and performance degradation can be occurred due to this non
(AWGN) and two fundamental fading channels for underwater orthogonality. So, we look for orthogonal signals [11].
acoustic systems as Rayleigh and Rician have been evaluated. Subsequently, the orthogonality between symbols can be
In section IV, a real analytical measurement is given which considered through evaluation of cross correlation coefficient.
predicts our simulated results. Finally, section V concludes Let Eb be the bit energy and defined as
this paper. s * (t ) Z Ts
0
Ts ☎0 Eb = s2i (t)dt, i = 0, 1 (5)
dt ✁
text 0
0
Decision
In this section, signal waveforms have been expressed ρ= (e∓jπ 2T t )∗ e±jπ 2T t dt (7)
Ts 0
mathematically. The orthogonality and Doppler effect on the
signals will be discussed, too. Using some straightforward manipulations, it can simplified
as
A. LFM s √
BTs
√
BTs
πx2 πx2
Z Z
1
|ρ| = √ ( sin dx)2 + ( cos dx)2 (8)
As we know, the chirp signals can be classified into two BTs 0 2 0 2
types; ”linear chirp”, in which the instantaneous frequency
decreases or increases with time in a linear form and ”non-
The above equation is composed of two famous integration
linear chirp”, in which the instantaneous frequency is a non
as Fresnel integrals. From (8), we can see ρ depends on
linear function of time. LFM probably is the most popular
spreading factor BT. As shown in Fig.2, the cross correlation
frequency modulation due to its low complexity. It is widely
coefficient decreases rapidly from 0 to 20 (secHz). For ρ ≤
used before in radar and sonar applications. Here, it is applied
0.1, we should choose BT more than 70 (secHz). This is why
to perform an underwater digital communication. Linear Up
LFM system requires a wide bandwidth for high data rate
and Down chirp can be expressed as
communication.
In case of Doppler effect, since the source and the receiver
B 2
Down :
s0 (t) = cos 2π(fc t − 2Ts
t ) 0 ≤ t ≤ Ts move with relative speed ν, the signal waveform suffers a
(1)
B 2
Up : s1 (t) = cos 2π(fc t + t ) 0 ≤ t ≤ Ts
2Ts
B. Trigonometric Chirp
−1
Generally, a pair of chirps can be written as following 10
Bit Error Probability
f0 (t) = cos 2π(fc t − Bθ(t)) 0 ≤ t ≤ Ts Trigonometric Chirp, BW= 400 Hz
−2
(10) 10 Trigonometric Chirp, BW= 500 Hz
Trigonometric Chirp, BW= 600 Hz
f1 (t) = cos 2π(fc t + Bθ(t)) 0 ≤ t ≤ Ts Linear Chirp, BW= 400 Hz
Linear Chirp, BW= 500 Hz
Linear Chirp, BW= 600 Hz
−3
10
The instantaneous frequency of a chirp as ”chirp rate
µ” which is defined by µ(t) = dθ(t) dt , denotes the type
of chirp signal. Now, it is necessary to express that the −4
10
trigonometric chirp signal has different types based on its
frequency spectrums. Here, we use full period trigonometric
−5
chirps whose frequency spectrum is a full period cosine or 10
−2 0 2 4 6 8
sine curve, just because of better performance in comparison Eb/N0 (dB)
−1 −1
10 10
Bit Error Probability
−3 −3
10 10
−4 −4
10 10
Trigonometric Chirp, Fd=50 Hz
Trigonometric Chirp, Fd= 150 Hz
−5 Trigonometric Chirp, Fd=150 Hz
10 −5 Trigonometric Chirp, Fd= 50 Hz
Linear Chirp, Fd=50 Hz 10
Linear Chirp, Fd= 50 Hz
Linear Chirp, Fd=150 Hz Linear Chirp, Fd= 150 Hz
−6 −6
10 10
−2 0 2 4 6 8 −2 0 2 4 6 8
Eb/N0 (dB) Eb/N0 (dB)
Fig. 4: Linear and Trigonometric chirp Bit Error Rate over Noisy and Doppler Fig. 6: Linear and Trigonometric chirp Bit Error Rate over Rayleigh fading Channel,
Channel, BT = 4 and fc = 8kHz. BT = 4 and fc = 8kHz.
0
10
Also, Fig.5 shows that how the performances of both linear Trigonometric Cirp, K=1
Trigonometric Chirp, K=2
and trigonometric chirp are affected by different Doppler shifts −1
10
Trigonometric Chirp, K=5
Linear Chirp, K=1
in a fixed SNR and BT. We can see that the full period
0
10
−4
10
−1
10 −5
10
Bit Error Probability
−6
−2 10
10 0 5 10 15 20
Eb/N0 (dB)
−3
Fig. 7: Linear and Trigonometric chirp Bit Error Rate over Rician Channel, BT = 4.
10
LFM Trigonometric 1
0.9
T (ms) BER T (ms) BER
0.8
0.7
10 0.0063 10 0
0.6
|h(t)|
20 0.0055 20 0 0.5
0.4
0.3
1 0.2
0.9 0.1
0.8 0
−2 0 2 4 6 8 10
0.7 Time Delay (ms)
0.6
(a) |h(t)|, BT = 4 and T = 20 ms
|h(t)|
0.5
1
0.4
0.9
0.3
0.8
0.2
0.7
0.1
0.6
0
|h(t)|
−2 0 2 4 6 8 10
0.5
Time Delay (ms)
0.4
(a) |h(t)|, BT = 4 and T = 20 ms
0.3
1 0.2
0.9 0.1
0.8 0
−2 0 2 4 6 8 10
0.7 Time Delay (ms)
0.6
(b) |h(t)|, BT = 4 and T = 10 ms
|h(t)|
0.5
0.4
Fig. 9: Channel Impulse Response of Water Tank in presence of Doppler effect
0.3
0.2
V. C ONCLUSION
0.1
TABLE II: Real UWA trials for LFM and Trigonometric when BT = 4, fc = 8kHz R EFERENCES
and ν = 2m/s
LFM Trigonometric [1] M. Stojanovic, “On the relationship between capacity and distance in
an underwater acoustic communication channel,” ACM SIGMOBILE
T (ms) BER T (ms) BER Mobile Computing and Communications Review, vol. 11, no. 4, pp. 34–
43, 2007.
[2] I. F. Akyildiz, D. Pompili, and T. Melodia, “State-of-the-art in protocol
10 0.2772 10 0.147
research for underwater acoustic sensor networks,” in Proceedings of
the 1st ACM international workshop on Underwater networks, pp. 7–
20 0.1941 20 0.0813 16, ACM, 2006.
Abstract— Secured storage in cloud computing has been one of enable public auditability for cloud data storage to allow users
the challenging areas in research. Searching and retrieval of to resort to a Third Party Auditor (TPA), who has expertise
outsourced encrypted data is very difficult due to the voluminous and capabilities to audit the outsourced data when needed [3].
data stored in cloud. Various strategies need to be devise to To avoid the leakage of the outsourced data to external parties,
eradicate unauthorized access, retrieval and use of such data. data encryption can be adopted to alleviate this privacy
Employing a secured scheme for an outsourced data is very
concerns. Encryption ensures data confidentiality, however
relevant in order to protect the confidentiality and privacy of the
data content. data search also becomes a challenge. Secure search over
In this work, we propose a Multi-Keyword Search employing encrypted data was first initiated by Song et al [4] by
Identity-based Encryption techniques (MKS-IDE) which enables proposing a cryptographic primitive concept called searchable
a data user to search for and retrieve encrypted files that has encryption which enables users to perform a keyword-based
been outsourced into the cloud by a data owner. The retrieved search on an encrypted data, just as on plaintext data. Wang et
encrypted files are ranked based on their relevance scores and al [5] were the first to define secure search over encrypted
the top-k relevant files are returned by the cloud to the data user. cloud data, however, further development made by
The data user then obtains a decryption key from the [6][7][8][9] and more, seeks to reduce computation and
appropriate data owner to decrypt his selected file. The security
storage cost and also enrich the category of search functions
requirement of our scheme is provably secure and the
performance of our scheme is also more efficient as compared to such as fuzzy keyword search, secure ranked multi-keyword
other PEKS schemes. search and similarity based search but these works are limited
to single-owner model. In 1984, a public key encryption
Keywords— Cloud Computing, Identity-Based Encryption scheme was introduced by Shamir [10]. The notion of identity
(IDE), Multi-Keyword Search, Privacy Preserving based cryptosystem is that the public key can be an arbitrary
string. Shamir's original motivation for identity-based
encryption was to simplify certificate management in e-mail
systems. Since the problem was posed in 1984 there have been
1. INTRODUCTION several proposals for IBE schemes [11][12]. Boneh and
The rapid growth in cloud computing is due to it wide usage Franklin in 2001, designed the first practical identity-based
and popularity. The tremendous increase in the data stored in cryptosystem [13]. Zhang et al [14] defined a multi-owner
cloud create security and privacy issues. These challenges are model for privacy preserving keyword search over encrypted
not only on how to store and manage these data but also on cloud data, however, their scheme did not use identity based
how to effectively and efficiently analyze these data to gain encryption. A similar system was proposed in [15] of which a
insightful knowledge in making smarter decision [1][2] and cryptographic techniques, query, response randomization and
also to ensure the data is secured enough. ranking capability was used. However, IDE was not used in
Cloud computing is a network-based environment that centers their system.
on sharing computations or resources. The numerous benefits In our work, we propose a multi-keyword search in cloud that
of cloud has accounted for different data owners outsourcing employs an identity-based encryption techniques (MKS-IDE).
their voluminous data. IT organizations have articulated The framework of our scheme and its security requirements are
concern about critical security issues that exist with the defined. We prove that our proposed scheme satisfies the
widespread implementation of cloud computing. ciphertext and trapdoor indistinguishability in the random
Security in cloud has been one of the most argued-about issues oracle [16][17]. This work adopt the ranking technique used in
in cloud computing making privacy preserving a very [14]. Finally, we demonstrate the advantage of our MKS-IDE
important mechanism in cloud computing. Outsourcing data scheme by comparing with previous PEKS schemes.
into the cloud is economically attractive for the cost and This paper is organized as follows: section II outlines the
complexity of long-term large-scale data storage, however, system model, threat model, and design goals. Section III states
data integrity and availability is not guaranteed. Thus, to the preliminaries, Section IV outlines the algorithms and the
efficiently verify the correctness of the outsourced cloud data security requirements of our scheme. We present a concrete
without the local copy of the data becomes a big challenge for description of our proposed MKS-IDE scheme in section V.
data storage security in cloud computing. Therefore, to fully Section VI has the security analysis. Section VII, we briefly
guarantee the security of the data and save the cloud user’s outline the privacy preserving ranking method which we
computation resources, it is of appropriately importance to
adopted from [14], section VIII gives the performance analysis Computational Diffie-Hellman (CDH) problem
of our scheme and we conclude in section IX.
4. ALGORITHMS AND SECURITY
2. SYSTEM AND THREAT MODEL AND DESIGN GOALS REQUIREMENTS
In this section, we describe the system model, threat model, We define the algorithms and security requirements for our
and design goals. MKS-IDE by modifying the ones in [8] [10] [18]
2.1 System Model 4.1 Algorithms
In our MKS-IDE scheme, we have three entities namely; data An MKS-IDE scheme consist of the following polynomial
owner, data user and cloud server. The data owner has a time randomized seven (7) algorithms:
collection of files to be outsourced to the cloud. To ensure i. Setup: A probabilistic algorithm which takes a
confidentiality, these files ought to be encrypted. For efficient security parameter l as input and generates a master
and adequate search operation, the data owner builds a secure secret / public key pair.
searchable index on keywords sets extracted from the files. ii. Key extract: This is a deterministic algorithm which
The data owner then outsource the encrypted files together takes as input a user’s identity, a master secret key
with their indexes. To facilitate keyword search over these and system parameters and returns a user secret key
encrypted files, a data user submits a hashed-keywords to the
skID .
data owner which is then used for trapdoor creation for the
iii. MKS-IDE: This is a probabilistic algorithm which
data user. Upon receiving the trapdoor, the data user submits it
takes a public key of the data owner, a set of
to the cloud. The cloud server then performs secure search
keywords in a document and system parameters and
over the encrypted indexes from various data owners and
replies with MKS-IDE searchable ciphertext index
return the corresponding set of files that matches same
I . The data owner encrypts each document with
patterns as that in the trapdoor. The cloud then rank these files
Symmetric-Key Encryption method using different
based on their relevance score and return the top-k important
keys for each document. The symmetric keys is then
files to the data user. The data user can then obtain a
encrypted with a Public Key Encryption which has
decryption key from the data owner to decrypt the file.
blinding capabilities.
2.2 Threat Model iv. Trapdoor generation: Given multiple keywords, data
In our threat model, we assume that the cloud server is honest user’s secret key and system parameters, this
but curious. The server follows our protocol but it keen to probabilistic algorithm generates the trapdoor Tw .
know the content of the encrypted files, keywords and the v. Test: This deterministic algorithm takes the MKS-
relevance scores. The data owner and data users who are IDE searchable ciphertext index I , trapdoor Tw and
authenticated by PKI are assumed to be trusted. system parameters. Upon receiving a query request
Tw from the user, the cloud server match all
2.3 Design Goals
keywords stored on it against the query request. It
In order to ensure privacy preserving, our MKS-IDE scheme then extracts and return the candidate files whose
should satisfy some security and performance goals. The indexes have same pattern as the query request.
proposed MKS-IDE should vi. Ranking: The cloud ranks the candidate file set
Enable multi-keyword search over different according to their relevance score and forward the
encrypted files from data owners. top-k relevant file to the data user. An order and
Must provide data user scalability. privacy preserving encoding scheme which encodes
Ensure that authenticated data users perform correct the relevance score to obtain the top-k search result
and appropriate search will be adopted into our MKS-IDE scheme.
vii. Retrieval: This algorithm takes a blinded encrypted
3. PRELIMINARIES symmetric key and returns a blinded symmetric key
for decryption of a document.
This section gives a brief review of the various concepts of
bilinear pairing, identity-based encryption and other related
mathematical problems used in this paper. Due to limited 4.2 Security Requirements
page, we omit the details of these concepts. Our MKS-IDE scheme must meet the following security
requirements: (1) ciphertext indistinguishability and
3.1 Bilinear Pairing
(2) trapdoor indistinquishability.
3.2 Identity-Based Encryption We define a security for Indistinguishability of Ciphertext
3.3 Other mathematically related problems and assumptions from Ciphertext of Chosen Keyword Attack (IND-CC-CKA)
3.3 Other mathematically related problems and assumptions. and trapdoor indistinquishability of the MKS-IDE scheme in
Two mathematical hard problems are used. Game 1 and 2 involve interactions between an adversary A
Bilinear Diffie-Hellman (BDH) problem and a challenger C. Since there is limited pages, we omit the
proof of both games.
b) Otherwise C generates a random coini {0,1} so that Trapdoor queries: Adversary A issues a trapdoor query for
Pr[coini 0]= for some that will be determined keyword IDi, wi , the Challenger C then access the
later. corresponding tuples IDi, Qi, bi, coini and
c) The challenger C picks a random value bi Zq* . IDi, wi, Qwi, xi in the H 1LIST and H 2 LIST respectively to
If coini 0 , C computes Qi biP G1* . computes the trapdoor Twi xiskIDi t where xi Zq* and
If coini 1 , C computes Qi biQID G1* . t H 2(wi ) for the adversary A.
d) The challenger C adds the tuple IDi, Qi, bi, coini a) Challenge: The adversary A sends ( IDC* ,W 0*W 1* ) to
to H 1LIST and returns H 1( IDi ) Qi to A. Note that the challenger C where W 0* and W 1* are two
challenged keywords. Upon receiving
Qi is uniform in G1* and is independent of A’s
( IDC* ,W 0*W 1* ) from A, the challenger C picks a
current view as required.
random value b {0,1} and access the corresponding
H2 queries: Similarly, at any time the adversary A can query tuple IDc* ,Wb* , Qwb* , x* in H 2 LIST to generate an
the random oracle H 2 . The Challenger C reacts to a query by
MKS-IDE ciphertext I * (U * ,V * ) (U * , R) , where
maintaining a list of tuple IDi, wi, Qwi, xi called the
H 2 LIST . The list is initially empty until the adversary makes a R {0,1}n is a random value. The restrictions
query. When the adversary A queries the oracle H 2 for remains that the adversary A did not queried a private
IDi, wi , The challenger C responds as follows: key extraction for IDc* and also for trapdoor query
a) If IDi, wi appears in the list H 2 LIST , C replies for W 0* and W 1* . The challenger C eventually sends
with H 2( IDi, wi) Qwi I * to the adversary A.
b) Phase 2: A continues with the request for key extract
b) Otherwise, the challenger C selects randomly xi Zq* queries adaptively for any identity IDC and the
and computes Qwi H 2(IDi, wi) xi.P . Finally, C trapdoor query for any keyword W from the
adds the tuple IDi, wi, Qwi, xi in the list H 2 LIST challenger C subject to the restriction that
and replies to the adversary A with IDc IDc* and w {W 0* or W 1*} .
H 2( IDi, wi) Qwi c) Guess: Finally, the adversary A outputs a guess of a
value b' {0,1} . A wins the game if b' b .
H3 queries: The adversary A can query the random oracle H 3
at any time. The challenger C responds to the query by By the assumption, the adversary A with a non-negligible
maintaining a list of tuple mi , ni called H 3 LIST which is probability 1 can distinguish the MKS-IDE ciphertext I *
initially empty. Responds to query is described as follows: under an adaptive chosen plaintext attack. Now, the challenger
The challenger C replies to the adversary’s queries to the C picks a tuple m* , n* in the H 3 LIST and outputs
oracle H 3 as indicated below: v* m* / e( H 2(wb), P) as the solution for the BDH instance
a) If mi appears in the list H 3 LIST , the challenger C
(P, aP, bP, cP) for a, b, c Zq* . It can be deduced that the
replies with H 3(mi ) ni
b) Otherwise, the challenger C select at random output m* / e( H 2(wb), P) is equal to e( P, P)abc . Observe that
ni {0,1}n and fix H 3(mi ) ni . The challenger C U aQID and m* (e(U , Ppub)e( H 2(wb), P)) . The challenger
then adds the pair mi, ni to the H 3 LIST and return C accesses the corresponding tuple IDc* ,Wb* , Qwb* , x* in
H 3(mi ) ni to the adversary A. the list H 3 LIST and compute
(e(aQID, cP) e(abP, cP) e( P, P) abc
Phase 1: Let IDi be a private key extraction query issued by
A. The Challenger C replies as follows: By adopting the similar technique used in [18][20], we
a) C runs the algorithm above for responding to H 1- compute the probability of our MKS-IDE scheme by
Queries to obtain a Qi G1* such that H 1( IDi ) Qi discussing the probability that the challenger C does not abort
and let IDi, Qi, bi, coini be the corresponding during the challenge phase. Suppose the adversary A makes
qR queries to the key extract query. In such an instance, the
tuple on the H 1LIST . If coini 1 , then C reports
probability that the challenger C does not abort in phase 1 or 2
failure and abort. The attack on MKS-IDE failed. qR
b) If coini 0 , C gets Qi biP and defines is ( ) and the probability that the challenger C does not
terminate during the challenged step is (1 ) . Therefore, the
skIDi biPpub G1* . Observe that skIDi cQi and
probability that the challenger C does not terminate during the
therefore skIDi is the private key associated to public qR qR
simulation is ( ) (1 ) . This value can be maximize at
key IDi . The Challenger then sends skIDi to A.
1 obtain:
abt 1 . By using abt , the probability that the
qR 1
1
e(Twb* H 2(wb), P) x e(abP, P)
1 1
challenger C does not abort is at least . Observe that Which implies that e(abP, P) e(Twb* H 2(wb), P) x . Hence
e(qR 1)
the Challenger can obtain abP x1 (Twb* H 2(wb)) , which
the probability analysis uses the same techniques as Coron’s
analysis of the Full Domain Hash in [21]. The Challenger C contradicts the CDH assumption.
2 1
outputs the correct D with the probability at least [21] Theorem 6.2 The proposed MKS-IDE scheme is semantically
qH 3 secure and satisfies the trapdoor indistinquishability against
where qH 3 represents the total number of requesting H 3 an adaptive chosen plaintext attack under the CDH
queries. Therefore, the Challenger C with a probability assumption in the random oracle.
2 1
can solve the BDH problem. This contradict
e(qR 1).qH 3 7. PRIVACY PRESERVING RANKED SEARCH
to the BDH assumption. Our scheme adopts the privacy preserving ranked search in
[14] . It is important to note that, after the retrieval of all the
Theorem 6.1: The proposed non-interactive searchable candidate files, the cloud server cannot return all the
encryption scheme (MKS-IDE) above is semantically secure undifferential files to the data user due to (1) acquisition of
against an adaptive chosen plaintext attack in the random excessive communication cost and overhead for the system if
oracle by satisfying the ciphertext indistinquishability under the cloud server decides to return all the candidate files. (2)
the BDH assumption. the data users may only be concerned with the top-k relevant
files that correspond to their queries. The scheme in [14]
6.1 Trapdoor Indistinguishability illustration an additive order preserving and privacy
preserving encoding scheme. It then uses the encoded
Lemma 6.2 In the random oracle model, we assume that if
relevance score to obtain the top-k search result.
there is an adversary A with a non-negligible advantage that
can break the trapdoor indistinguishability of the proposed
MKS-IDE scheme under the adaptive chosen keyword attack, 8. PERFORMANCE ANALYSIS AND COMPARISON
then there exist a challenger C with a non-negligible
advantage who can solve the computational CBH problem. Adopting similar strategy used in [18] we analyze the
performance of our MKS-IDE scheme. We compare our
Proof: The challenger C receives a CDH instance of MKS-IDE scheme with some previously proposed PEKS
schemes by concentrating more on some time consuming
(P, aP, bP) for a, b Zq* where q is the order of G1 and G 2 . operations defined as follows:
By interacting with the adversary A, the challenger C returns
the CDH solution abP in Game 2. Since there is limited TGe : The execution time of a bilinear map operation
pages, we omit the proof for trapdoor indistinguishability e : G1 G1 G2
since it’s similar as defined in lemma 6.1. TGmul : The execution time for scalar multiplication
operation in G1
It is assumed that the adversary A with a non-negligible
TGH : The execution time for map-to-point hash
advantage can differentiate the trapdoor Twb* under the
*
function, thus H 1, H 2, H 3 :{0,1}* G1
adaptive chosen keyword attack. Meaning the trapdoor Twb Tinv : The execution time of a modular inverse
satisfy the equation e(Twb* , P) V ' where operation in Zq
V ' e(U , Ppub).e( H 2(wb), P) and Let U xQID . We can then The most time consuming operation is the execution time for
TABLE 1: Comparison between our IDRMKS and previously proposed dPEKS schemes
Computational cost for ciphertext 3TGmul TGe 2TGmul TGe 2TGmul TGe 3TGmul 2TGe TGmul
generation (1 keywords) 2TGH TGH TGH 3TGH 2TGH
Computational cost for trapdoor 3TGmul 2TGH 3TGmul 2TGH 3TGmul 2TGH 2TGmul TGH TGmul TGH
generation Tinv Tinv Tinv
Computational cost for test 3TGe TGe 2TGmul TGe 2TGmul 2TGe TGmul TGe 2TGmul
ISBN: 978-1-941968-47-5 ©2018 SDIWC TGH TGH Tinv Tinv17
Proceedings of the Fifth International Conference on Technological Advances in Electrical, Electronics and Computer Engineering (TAEECE2018), Turkey, 2018
bilinear map operation TGe as compared to the other [3] C. Wang, Q. Wang, K. Ren, and W. Lou, “Privacy-preserving
public auditing for data storage security in cloud computing,” Proc.
operations stated above. In [22], the performance simulation - IEEE INFOCOM, 2010.
results show that TGe 2.5TGmul . We therefore analyze the [4] D. Wagner, A. Perrig, D. X. Song, D. Wagner, and A. Perrig,
performance of our MKS-IDE scheme for each phase. In the “Practical techniques for searches on encrypted data,” in Proceeding
MKS-IDE ciphertext generation phase, it required 2000 IEEE Symposium on Security and Privacy, 2000, pp. 44–55.
[5] P. Golle, J. Staddon, and B. Waters, “Secure conjunctive keyword
2TGe TGmul (n 1)TGH to generate an MKS-IDE search over encrypted data,” Appl. Cryptogr. Netw. Secur., vol.
ciphertext, note that n represents the total number of 3089, pp. 31–45, 2004.
[6] R. Li, Z. Xu, W. Kang, K. C. Yow, and C. Z. Xu, “Efficient multi-
keywords. TGmul TGH is essential to generate the trapdoor keyword ranked query over encrypted data in cloud computing,”
Tw in the trapdoor generation phase, Futur. Gener. Comput. Syst., vol. 30, no. 1, pp. 179–190, 2014.
[7] J. Li, Q. Wang, C. Wang, N. Cao, K. Ren, and W. Lou, “Fuzzy
Table 1 above outlines the comparison between our MKS-IDE Keyword Search over Encrypted Data in Cloud Computing,” 2010
scheme to other dPEKS schemes [23][24][25] ] in terms of Proc. IEEE INFOCOM, pp. 1–5, 2010.
[8] B. Wang, S. Yu, W. Lou, and Y. T. Hou, “Privacy-preserving multi-
public key setting and performance. From the table in [18], it keyword fuzzy search over encrypted data in the cloud,” in
can be observed that conjunctive keywords is not supported Proceedings - IEEE INFOCOM, 2014, pp. 2112–2120.
by the schemes in [23][25] and they require [9] M. Chuah and W. Hu, “Privacy-aware bedtree based solution for
fuzzy multi-keyword search over encrypted data,” in Proceedings -
nTGe (n 1)TGmul nTGH for their ciphertext generation. We International Conference on Distributed Computing Systems, 2011,
can therefore conclude from table 1 above that our MKS-IDE pp. 273–281.
scheme is more efficient than the schemes in [23][24][25] with [10] A. Shamir, “Identity-Based Cryptosystems and Signature Schemes,”
Adv. Cryptol., vol. 196, pp. 47–53, 1985.
respect to ciphertext generation especially when n is [11] U. Maurer and Y. Yacobi, “Non-interactive public-key
sufficiently large and also efficient in the trapdoor generation. cryptography,” Adv. Cryptology—EUROCRYPT’91, pp. 498–507,
Since our MKS-IDE scheme is based on ID system, the load 1991.
of certificate management associated with the other schemes [12] L. Problem, “An ID-Based Cryptosystem Based on the Discrete,”
vol. 7, no. 4, 1989.
which are based on pairing-based public key system are [13] D. Boneh and M. Franklin, “Identity-Based Encryption from the
eradicated. Although the scheme in [25] is also based on ID- Weil Pairing,” SIAM J. Comput., vol. 32, no. 3, pp. 586–615, 2003.
system, however in the ciphertext and trapdoor generation, our [14] W. Zhang, Y. Lin, S. Xiao, J. Wu, and S. Zhou, “Privacy Preserving
MKS-IDE is more efficient than [25]. Ranked Multi-Keyword Search for Multiple Data Owners in Cloud
Computing,” IEEE Trans. Comput., vol. 65, no. 5, pp. 1566–1577,
2016.
[15] P. D. Mgmt, C. A. Secondary, and C. Author, “Distributed and
9. CONCLUSION Parallel Databases An Efficient Privacy-Preserving Multi-Keyword
In this paper we proposed an MKS-IDE scheme which Search over Encrypted Cloud Data with.”
[16] R. Canetti, O. Goldreich, and S. Halevi, “The random oracle
supports conjunctive keywords. We defined the framework methodology, revisited,” J. ACM, vol. 51, no. 4, pp. 557–594, 2004.
and the security requirements for our MKS-IDE scheme and [17] M. Bellare, “R a n d o m Oracles are Practical : A Paradigm for
when we compared our scheme to previous dPEKS scheme, Designing Efficient P r o t o c o l s,” pp. 62–73, 1993.
the performance of our scheme is more efficient in both the [18] T.-Y. Wu, T.-T. Tsai, and Y.-M. Tseng, “Efficient searchable ID-
based encryption with a designated server,” Ann. Telecommun. -
ciphertext and trapdoor generation phase. Our ID-based Ann. Des Télécommunications, vol. 69, no. 7–8, pp. 391–402, 2014.
system also has an advantage of eliminating certificate [19] C. Orencik and E. Savas, “Efficient and secure ranked multi-
management associated with PKI. We also demonstrated that keyword search on encrypted cloud data,” Proc. 2012 Jt.
our MKS-IDE scheme possesses the ciphertext EDBT/ICDT Work., pp. 186–195, 2012.
[20] Y. M. Tseng and T. T. Tsai, “Efficient revocable ID-based
indistinquishability and trapdoor indistinguishability under encryption with a public channel,” Comput. J., vol. 55, no. 4, pp.
BDH and CDH assumptions respectively. 475–486, 2012.
Our future work will seek to implement our MKS-IDE scheme [21] J.-S. Coron, “On the Exact Security of Full Domain Hash,”
in a hybrid cloud and also to investigate our MKS-IDE CRYPTO 2000 Adv. Cryptol., pp. 229–235, 2000.
[22] T.-Y. Wu and Y.-M. Tseng, “An ID-Based Mutual Authentication
scheme without random oracle with a pairing free algorithm to and Key Exchange Protocol for Low-Power Mobile Devices,”
determine how feasible it will be in cloud. Comput. J., vol. 53, no. 7, pp. 1062–1070, 2009.
[23] C. Hu and P. Liu, “An enhanced searchable public key encryption
scheme with a designated tester and its extensions,” J. Comput., vol.
REFERENCES 7, no. 3, pp. 716–723, 2012.
[24] Y. H. Hwang and P. J. Lee, “Public key encryption with conjunctive
[1] I. Abaker, T. Hashem, I. Yaqoob, N. Badrul, S. Mokhtar, A. Gani, keyword search and its extension to a multi-user system,” Lect.
and S. Ullah, “The rise of ‘ big data ’ on cloud computing : Review Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect.
and open research issues,” Inf. Syst., vol. 47, pp. 98–115, 2015. Notes Bioinformatics), vol. 4575 LNCS, pp. 2–22, 2007.
[2] L. Ballard, S. Kamara, and F. Monrose, “Achieving efficient [25] H. S. Rhee, J. H. Park, W. Susilo, and D. H. Lee, “Trapdoor security
conjunctive keyword searches over encrypted data,” in Lecture in a searchable public-key encryption scheme with a designated
Notes in Computer Science (including subseries Lecture Notes in tester,” J. Syst. Softw., vol. 83, no. 5, pp. 763–771, 2010.
Artificial Intelligence and Lecture Notes in Bioinformatics), 2005,
vol. 3783 LNCS, pp. 414–426.
The written code and the used libraries are expressed with 68
dots in the face area using the frames to locate the eye, Figure 6 Example of opening and closure of an eye.
eyebrow, nose, mouth and chin lines of the driver (Figure 3).
• Logitech Speakers: Speakers are used to provide audio Example of the tests done in the lab are shown in Figure 9 and
alarm. It works with 5V. Figure 10. Figure 9 shows eye detection of closed eyes where
• Raspberry pi chamber/case and fan: The case is used to the opening ratio between upper and lower parts is smaller
protect the system from external environment, and a fan is than threshold. Alarm will be activated.
used to prevent the system from overheating due to its long
usage.
Software used in the study:
• Python Language: The programming language used in the
project.
• Opencv and Dlib Libraries: Independent libraries used in
this project to develop real-time computer vision applications.
• Haar Cascade Classifier: The classifier used in the project
identifies and implements the screening method of the human
face detection.
The above mentioned are the main software. Additionally;
time, imutils, video, numpy, argparse libraries are also used.
References
[1] WHO. Road Safety: Basic Facts; 2016
(http://www.who.int/violence_injury_prevention/publications/
road_traffic/Road_safety_media_brief_full_document.pdf).
[2] Monthly Traffic Accidents report in Turkey. April 2018.
http://www.trafik.gov.tr/SiteAssets/istatistik/nisan18.pdf
[3] D E King. “Dlib-ml: A Machine Learning Toolkit”. Journal of
Machine Learning Research, vol 10, 2009.
[4] P Viola, M Jones, “Rapid Object Detection Using A Boosted
Cascade of Simple Features”, In proceedings of Computer
Vision and Pattern Recognition (CVPR01), pp. 511-518, 2001.
[5] V Tengshe, V Tengshe, “Drowsy Driving Alarm System”,
2005, US Patent US20060214807A1, 2005.
[6] A Sahayadhas, K Sundaraj, M Murugappan. “Detecting Driver
Drowsiness Based on Sensors: A Review”. Sensors, vol.12,
no.12, pp. 16937-16953. 2012.
[7] R Grace, S Steward, “Drowsy Driver Monitor and Warning
System”, Proceedings of the First International Driving
Symposium on Human Factors in Driver Assessment, Training
and Vehicle Design. USA, 2001, pp. 64-69.
Figure 12 Eye detection of open eyes (inside the car) [8] E E Galarza, FD Egas, F M Silva, P M Velasco, E D Galarza,
“Real Time Driver Drowsiness Detection Based on Driver’s
Face Image Behavior Using a System of Human Computer
Interaction Implemented in a Smartphone”, Proceedings of the
International Conference on Information Technology &
Systems (ICITS 2018), pp. 563-572, 2018.
IV. Conclusions
SUDAD J ASHAJ
Gaziantep University
sudad_80@yahoo.com
Prof. Dr. Ergun ERÇELEBİ
Gaziantep university
ergun.ercelebi@gmail.com
microprocessor and Linux operation system. aggregated by the PC server and appropriately
The proposed system has the voice prompt, compare them for making statistical analysis
remote control greenhouse device and remote info.
monitor. The test of system shows that it is
stable, and it can be developing to use as an Suresh et al. (2016) [7], developed an
authorized devise to be maintains that is automatic control system for lighting in order
appropriate for DAQ and as device control for to got an effective use of energy. They also
several kinds of agricultural sites. have presented mobility and remote command
performance to system by using Android
2 LITERATURE SURVEY mobile App through Bluetooth to control the
lighting based upon the voice command
This section discussed the in short the several
literatures presented related to Smart Home
and Energy Management System 3 PROPOSED SYSTEM STRUCTURES
A. Software design:
NO YES
Accuracy >90 OR
time < 20 second
Any more
codes
Fig.3. Raspberry pi connection with
sensors
5 CONCLUSION
1- The proposed system can process, display, Instrumentation Engineering. vol.6, Issue 4, pp.
and store the data in both way (as DAQ 3162- 3169. April 2017.
system or dedicate system). [9] Raspberry official Site, Link:
2- The benefit of utilizing a Raspberry PI for https://www.raspberrypi.org/.
DAQ is that a Raspberry PI has the
flexibility to adjust the changing [10] J. R. C. León, R. F. Martínez-González, A. M.
requirements and its low cost. Medinay and L. A. Peralta-Pelaez. “Raspberry Pi
and Arduino Uno Working Together as A Basic
3- The design and constricted of this model Meteorological Station,” International Journal of
can be done in highest accuracy with time Computer Science & Information Technology
that consumed. (IJCSIT) vol.9, no.5, pp.97-104, October 2017.