Sie sind auf Seite 1von 7

58 (IJCNS) International Journal of Computer and Network Security,

Vol. 2, No. 4, April 2010

Cluster Management using Cluster Size Ratio in Ad


Hoc Networks
D K L V Chandra Mouly, Ch D V Subba Rao and M M Naidu
Department of Computer Science and Engineering
S V University College of Engineering, Tirupati - 517502, India.
dklvcm.mtech@gmail.com, subbarao_chdv@hotmail.com

Abstract: Cluster Management using Cluster Size Ratio delivering messages in a decentralized environment where
(CMCSR) is a completely distributed algorithm for partitioning network topology fluctuates is not a welldefined problem
a given set of mobile nodes into clusters. The proposed [1].
algorithm tries to reduce the amount of computational and
information overhead while maintaining a stable cluster 1.2 Clustering in Ad Hoc Networks
formation. It constructs and maintains a backbone topology A wireless ad hoc network consists of nodes that move freely
based on a minimal dominating set (MDS) of the network.
and communicate with each other using wireless links. Ad-
According to this algorithm, each node determines the
membership in the MDS for itself and its one-hop neighbors
hoc networks do not use specialized routers for path
based on one-hop neighbor information that is disseminated discovery and traffic routing. One way to support efficient
among neighboring nodes using willingness and priority communication between nodes is to develop wireless
information of the nodes. The algorithm then ensures that the backbone architecture; this means that certain nodes must be
members of the MDS are connected into a connected selected to form the backbone. Over time, the backbone
dominating set (CDS), which can be used to form the backbone must change to reflect the changes in the network topology
infrastructure of the communication network to facilitate as nodes move around. The algorithm that selects the
routing. The algorithm outperforms the existing algorithms with members of the backbone should naturally be fast, but also
respect to stability. Load balancing the cluster heads using the should require as little communication between nodes as
cluster size ratio is the heuristic used in this algorithm. possible, since mobile nodes are often powered by batteries.
One way to solve this problem is to group the nodes into
1. Introduction clusters, where one node in each cluster functions as cluster
head, responsible for routing [2].
This section discusses elementary issues of ad hoc networks
and benefits of clustering. 1.3 Benefits of clustering
Ad-hoc networks are suited for use in situations where an
1.1 Ad Hoc Networks infrastructure is unavailable or to deploy one is not cost
In the next generation of wireless communication systems, effective. One of many possible uses of mobile ad-hoc
there will be a need for the rapid deployment of independent networks is in some business environments, where the need
mobile users. Significant examples include establishing for collaborative computing might be more important
survivable, efficient, dynamic communication for outside the office environment than inside, such as in
emergency/ rescue operations, disaster relief efforts, and business meeting outside the office to brief clients on a
military networks. Such network scenarios cannot rely on given assignment.
centralized and organized connectivity, and can be Mobile ad-hoc networks allow the construction of flexible
conceived as applications of ad hoc networks. An ad hoc and adaptive networks with no fixed infrastructure. These
network is an autonomous collection of mobile users that networks are expected to play an important role in the future
communicate over relatively bandwidth constrained wireless wireless generation. Future wireless technology will require
links. Since the nodes are mobile, the network topology may highly-adaptive mobile networking technology to effectively
change rapidly and unpredictably over time. The network is manage multi-hop ad-hoc network clusters, which will not
decentralized, where all network activity including only operate autonomously but also will be able to attach at
discovering the topology and delivering messages will be some point to the fixed networks.
taken care by the nodes, i.e., routing functionality will be
incorporated into mobile nodes. 2. Literature Review

The set of applications for ad hoc networks is diverse, This section emphasizes some of the past clustering
ranging from small, static networks that are constrained by techniques.
power sources, to large-scale, mobile, highly dynamic
networks. The design of network protocols for these 2.1 Types of Topology Management
networks is a complex issue. Regardless of the application, There are two approaches to topology management in ad
ad hoc networks need efficient distributed algorithms to hoc networks:
determine network organization, link scheduling, and
• Power control.
routing. However, determining viable routing paths and
• Hierarchical topology organization.
(IJCNS) International Journal of Computer and Network Security, 59
Vol. 2, No. 4, April 2010

2.1.1 Power Control C. Reactive Clustering


Different Power control mechanisms adjust the power on a It is on-demand clustering algorithm. There is no periodic
per-node basis, so that one-hop neighbor connectivity is exchange of clustering information in the network. Instead,
balanced and overall network connectivity is ensured [4, 5, whenever there is data traffic, cluster related information is
6]. Li [7] proved that network connectivity is minimally piggybacked in outgoing data packets and extracted out of
maintained as long as the decreased power level keeps at received packets.
least one neighbor remaining connected at every 2π/3 to D. Proactive Clustering
5π/6 angular separation. Ramanathan [7] proposed to Some are proactive clustering algorithms, which require
incrementally adjust nodes power levels so as to keep periodic broadcast of cluster-related information. SPAN [13]
network connectivity at each step topologies derived from adaptively elects coordinators according to the remaining
power-control schemes often result in unidirectional links energy and the number of pairs of neighbors a node can
that create harmful interference due to the different connect.
transmission ranges among one-hop neighbors [9]. The
dependencies on volatile information in mobile networks, 2.3 Scope for Present Work
such as node locations [4], signal strength or angular
positions [8] also contribute to the instability of topology The efficiency of a communication network depends not
control algorithms based on power control. only on its control protocols, but also on its topology. Our
work i.e. CMCSR proposes a distributed topology
2.1.2 Hierarchical topology control management algorithm that constructs and maintains a
This approach to topology control is often called clustering, backbone topology based on a Minimal Dominating Set
and consists of selecting a set of cluster heads in a way that (MDS) of the network. Without topology management each
every node is associated with a cluster head, and cluster and every node should maintain the routing information for
heads are connected with one another directly or by means all the nodes they need. By using topology management a
of gateways, so that the union of gateways and cluster heads subset of nodes are selected called cluster heads and each
constitute a connected backbone [10, 14, 15]. Once elected, cluster head performs the routing work for its members.
the cluster heads and the gateways helps to reduce the
complexity of maintaining topology information, and can 3. System Model
simplify such essential functions as routing, bandwidth
allocation, channel access, power control or virtual circuit 3.1 Assumptions
support. For clustering to be effective, the links and nodes This work assumes that an ad hoc network comprises a
that are part of the backbone (i.e., cluster heads, gateways, group of mobile nodes communicating through a common
and the links that connect them) must be close to minimum broadcast channel using omni-directional antennas with the
and must also be connected. same transmission range. The topology of an ad hoc
network is thus presented by an undirected graph G = (V,E),
2.2 TYPES OF CLUSTER HEAD ELECTIONS where V is the set of network nodes, and E ⊆ U * V is the
Cluster heads can be elected in four ways. set of links between nodes. The existence of a link (u, v) ∈ E
• Deterministic Clustering also means (v; u) ∈ E, and that nodes u and v are within the
packet-reception range of each other, in which case u and v
• Non-Deterministic Clustering
are called one-hop neighbors of each other. The set of one-
• Reactive Clustering
hop neighbors of a node i is denoted by Ni. Two nodes that
• Proactive Clustering
are not connected but share at least one common one-hop
A. Deterministic Clustering neighbor are called two-hop neighbor of each other.
Deterministic clustering can determine the cluster heads in a
single round. Different heuristics have been used to form Each node has one unique identifier, and all transmissions
clusters and to elect cluster heads. Several approaches [12] are omni directional with the same transmission range. The
utilized the node identifiers to elect the cluster heads within nodes move with constant mobility. The energy is decreased
one or multiple hops. linearly. Different types of nodes consume energy at
different rates. We ignore the energy consumed due to local
B. Non-Deterministic Clustering computations, but assume that the energy consumption rate
is only dependent on the type of the node. A host consumes
In non-deterministic clustering, negotiations are used.
0.6% of the total energy per minute in these algorithms;
Negotiations require multiple incremental steps, and may
whereas a cluster head consumes 3%.
incur an election jitter during the process, because of the
lack of consensus about the nodes being elected as the
3.2 Model
cluster heads. Examples of this approach are the “core”
extraction algorithm [13] and the spanning tree algorithm In an ad hoc network all nodes are alike and all are mobile.
[14]. SPAN [13] allows a node to delay the announcement There are no base stations to coordinate the activities of
of becoming a cluster head for random amounts of time to subsets of nodes. Therefore, all the nodes have to
attempt to attain minimum conflicts between cluster heads collectively make decisions. All communication is over
in its one-hop neighborhood. wireless links. A wireless link can be established between a
pair of nodes only if they are within wireless range of each
60 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 4, April 2010

other. We will only consider bidirectional links. It is priorities that consists of two components: (a) Neighboring
assumed the MAC layer will mask unidirectional links and Nodes, (b) Willingness value assigned to a node as a
pass only bidirectional links. Beacons could be used to function of its mobility and energy level.
determine the presence of neighboring nodes. After the We denote the willingness value of node i by Wi, the speed
absence of some number of successive beacons from a of node i by a scalar Mi that ranges from 0 to 1 meters per
neighboring node, it is concluded that the node is no longer second, and the remaining energy on node i as Ei in the
a neighbor. Two nodes that have a wireless link will, range of 0 and 1. The willingness Wi is a function that
henceforth, be said to be one wireless hop away from each should be defined according to the following criteria:
other. They are also said to be immediate neighbors.
Communication between nodes is over a single shared 1. To enhance survivability, each node should have the
channel. responsibility of serving as a cluster head with some
nonzero probability determined by its willingness value.
In ad hoc networks the nodes within each neighborhood are 2. To facilitate with the stability and the frequency with
not known a priori. The individual cluster may transition to which cluster head elections must take place, the
spatial TDMA for inter-cluster and intra-cluster willingness value of a node should remain constant as long
communication. All nodes broadcast their node identity as the variation of the speed and energy level of the node do
periodically to maintain neighborhood integrity. Due to not exceed some threshold values.
mobility, a node’s neighborhood changes with time. As the 3. To avoid electing cluster heads that quickly lose
mobility of nodes may not be predictable, changes in connectivity with their neighbors after being elected, the
network topology over time are arbitrary. However, nodes willingness value of a node should decrease drastically after
may not be aware of changes in their neighborhood. the mobility of the node exceeds a given value.
Therefore, clusters and cluster heads must be updated 4. To prolong the battery life of a node, its willingness
frequently to maintain accurate network topology. value should decrease drastically after the remaining energy
of the node drops below the given level.
3.2.1 Attributes of a node
Willingness value (Wi) is as specified below:
The attributes of a node and their functionality are as given
Wi = 2log2(Ei+.9)log2(Mi+2)
in Table I.
Here the constants 0.9 and 2 in Eq. (1) eliminate the
Table 1: Attributes of a node and their functionality
boundary conditions in the logarithmic operations. The
ATTRIBUTE FUNCTION logarithmic operations on the speed and the remaining
energy values render higher willingness values in the high
ID Unique name given to node
energy and low speed field, while giving close to zero values
ENERGY The capacity to work in
in the low energy and high-speed region.
MOBILITY The speed of the node when it
is moving Priority value (Pi) is a function of no.of neighbors and
WILLINGNESS How much the node is willing willingness
to be a cluster head i.prio = 2log2(Wi)/n
PRIORITY Has the priority among other Figure 1 illustrates the effect of the two factors on the
nodes to became a cluster head priority values. From the Figure 2 we can conclude that the
CLUSTER SIZE Cluster size ratio which it is priority is directly proportional to the willingness value and
having number of neighbors.
TYPE Whether it is cluster head or
gateway or door way or
member
NEIGHBORS Number of one – hop neighbors
1

3.2.2 Computing Priorities of Nodes 0.9


0.8
Given that cluster heads provide the backbone for a number 0.7
0.6
of network control functions, their energy consumption is Priority 0.5
more pronounced than that of ordinary hosts. Low-energy 0.4 10
nodes must try to avoid serving as cluster heads to save 0.3
7
0.2
energy. However, to balance the load of serving as cluster 4
0.1
heads, every node should take the responsibility of serving 0 1 Number of
as a cluster head for some period of time with some 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Neighbors

likelihood. Furthermore, node mobility has to be considered Willingness

in cluster head elections. To take into account the mobility


and energy levels of nodes in their election, we define the
two-hop neighbor information needed to assign node Figure 1. Priority Graph
(IJCNS) International Journal of Computer and Network Security, 61
Vol. 2, No. 4, April 2010

The objective is to develop an enhancement for existing


heuristics to provide a contiguous balance of loading on the
3.2.3 MDS and Cluster Head Election
elected cluster heads. Once a node is elected as cluster head
The approach to establishing a minimal dominating set it is desirable for it to stay as a cluster head up to some
(MDS) is based on three key observations First, using maximum specified amount of time, or budget. The budget
negotiations among nodes to establish which nodes should is a user defined constraint placed on the heuristic and can
begin the MDS incurs substantial overhead when nodes be modified to meet the unique characteristics of the system,
move around and the quality of links changes frequently. i.e., the battery life of individual nodes. Some of the goals of
Hence, nodes should be allowed to make MDS membership the heuristic are:
decisions based on local information. Second, because in an 1. Minimize the number and size of the data structures
MDS every node is one hop away from a cluster head, the required to implement the heuristic,
local information needed at any node needs to include only 2. Extend the cluster head duration budget based on an input
nodes that are one and two hops away from the node itself. parameter,
Third, having too many cluster heads around the same set of 3. Allow every node equal opportunity to become a cluster
nodes does not lead to an MDS. Hence, to attain a selection head in time,
of nodes to the MDS without negotiation, nodes should rank 4. Maximize the stability in the network.
one another using the two-hop neighborhood information
they need. Based on the above, the approach adopted in Data Structures
CMCSR consists of each node communicating to its The data structures necessary for the heuristic consist of one
neighbor’s information about all its two-hop neighbors. local variable: Physical ID (PID). The PID is the initial id
Using this information, each node computes a priority for given and is unique for each individual node. However, this
each node in its two-hop neighborhood, such that no two changes with time to represent the elect ability of a node.
nodes can have the same priority at the same instant of time.
A node can become cluster head if the node has highest Basic Idea
priority in its two hop neighborhood. The node id load heuristic operates on the principle of load
balancing. That is, the ids of each non-cluster head node
3.2.4 Connected Dominating Set Election cycles through the queue at a rate of 1 unit per run of the
load-balancing heuristic. Each node has a minimum value
The CDS [4] of a network topology is constructed in two
of 0 and a maximum value of Max_Cluster Size. Upon
steps. In the first step, if two cluster heads in the MDS are
reaching Max_Cluster Size a node will rotate to a value of 0
separated by three hops and there are no other cluster heads
on the next cluster election heuristic run. As the cluster
between them, a node with the highest priority on the
election heuristics run they will use the priorities to
shortest paths between the two cluster heads is elected as a
determine the cluster heads of the network. A cluster head
doorway, and is added to the CDS. Therefore, the addition
will maintain this value until it has exhausted its cluster
of a doorway brings the connected components in which the
head duration budget. At this point it will set its work to 0,
two cluster heads reside one hop closer. In the second step,
i.e., less than any other node, and become a normal node.
if two cluster heads or one cluster head and one doorway
node are only two hops away and there are no other cluster
heads between them, one of the nodes between them with 4. Performance Evaluation
the highest priority becomes a gateway to connect cluster We have conducted simulation experiments to evaluate the
head to cluster head or doorway to cluster head. After these performance of the proposed heuristic i.e. CMCSR. These
steps, the CDS is formed. simulation results were then compared against Topology
CDS is constructed in two steps Management by Priority Ordering TMPO [15]. We assumed
• Selecting doorway a variety of systems running with 10, 20, 40, 60, 80 and 100
• Selecting gateway nodes to simulate ad hoc networks with varying levels of
node density. Two nodes are said to have a wireless link
(a) Selecting Doorway between them if they are within communication range of
Node i can become a doorway for cluster heads n and j if the each other. Additionally, the span of a cluster, i.e. the
following conditions are satisfied. i) If cluster n and j are maximum number of wireless hops between a node and its
not two hops away. ii) There is no other cluster head m on cluster head (d) was set to 2. The entire simulation was
the shortest path between n and j. iii) There is no other node conducted in a 1150 * 1150 unit region. Initially, each node
m with higher priority than node i. was assigned a unique node id and (x, y) coordinates within
the region. The nodes were then allowed to move at random
Selecting gateway in any direction at a speed of not greater than half of the
wireless range of a node per second. The simulation range is
Node i can become a gateway for cluster head n and j if the
set to 2000 seconds, and the network was sampled for every
following conditions are satisfied.
2 seconds. At each sample time the proposed cluster size
i) If there is no cluster head or doorway between n and j.
ratio and cluster election heuristic was run to determine
ii) If there is no node with higher priority than node i.
cluster heads and their associated clusters. Each simulation
run for 2000 seconds measures several performance metrics.
3.2.5 Computing Cluster Size ratio
The main simulation metric measured was Cluster head
62 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 4, April 2010

Duration, which provides a basis for evaluating the shaped line indicates the cluster head duration with load i..e.
performance of the proposed load-balancing heuristic. incase of CMCSR.
For the purposes of these simulations we have set the cluster
head budget to be a function of the maximum amount of
4.2 Nodes Vs Number of Cluster Heads
work it performs That is, once a node becomes a cluster
head it will remain a cluster head until it has exhausted its Figure 3 shows the graph for the average number of cluster
maximum work load, or until it loses out to another cluster heads formed during the topology management. The
head based on the rules of the cluster election heuristic. topology management is executed for 1800 seconds for each
of x nodes and the values are noted. The diamond shaped
The proposed CMCSR algorithm makes a noticeable line indicates the number of cluster head formed during
difference in the cluster head duration (ranging from 4% to topology management without load (TMPO). Second the
28%). This shows that the load-balancing heuristics topology management executed for the 600 seconds for each
generates longer cluster head durations; it will also produce x nodes and the values are noted. Totally the program is
much tighter and more deterministic responses (stability). executed for 18000 seconds. The square shaped line
These results are not surprising. Therefore, once a cluster indicates the cluster head formed during topology
head is elected it continues as cluster head for a maximum management with load (CMCSR).
of the programmed budget. This will provide the longer
cluster head durations that we see. The cluster size ratio 16

heuristic is continuously rotating, moving ordinary nodes


14
into the position of becoming a cluster head. Therefore, once
a cluster head budget is exceeded, a different cluster head is 12
elected and the process repeats. This provides the cluster
size ratio effect of distributing the responsibility of being a No.of Clusters
10

cluster head among all nodes. We present below three


8
graphs for our simulation results. First one is the average
cluster head duration. Second one is the average number of 6
cluster head. And finally the improvement graph for the
cluster head duration. 4

2
4.1 Nodes Vs Cluster Head Duration
0
Figure 2 shows the graph for the average cluster head 10 20 30 40 50 60 70 80 90 100

duration. X-axis takes the number of nodes and y-axis No.of Nodes

shows the cluster head duration in seconds. The topology TMPO CMCSR
management is executed for 1800 seconds for each x nodes
Figure 3. Average no. of clusters
and the values are noted. Totally the program is executed for
18000 seconds. The diamond shaped line indicates the
cluster head duration without load i.e. incase of TMPO.
25
30

25
20
Cluster Head Duration(in Sec)

20

15
15

10
10

5
0
10 20 30 40 50 60 70 80 90 100
No. of Nodes
0
TMPO CMCSR
900 1800 2700 3600
Syst em Execut ed(Sec)
Figure 2. Average Cluster head duration Vs no. of nodes TMPO CMCSR

Second the topology management is executed for the 600 Figure 4. Average Cluster head duration
seconds for each x nodes and the values are noted. Totally
the program is executed for 18000 seconds. The square
(IJCNS) International Journal of Computer and Network Security, 63
Vol. 2, No. 4, April 2010

4.3 Improvement in Cluster head Duration Sciences Research 2(6): 336-340, INSInet Publication,
2006.
Figure 4 shows the graph for the average cluster head
[4] L. Hu. “Topology Control for Multihop Packet Radio
duration. X-axis takes system executed in seconds and y-
Networks,”. IEEE Transactions on Communications,
axis shows the average cluster head duration in seconds.
41(10), Oct. 1993.
The below graph is constructed under the following
[5] S. Narayanaswamy, V. Kawadia, R. S. Sreenivas, and
conditions. Both TMPO and CMCSR is run for 900 sec,
P. R. Kumar, “Power Control in Ad-Hoc Networks:
1800 sec, 2700 sec and 3600 sec by taking total number of
Theory, Architecture, Algorithm and Implementation of
nodes into account as 50. The diamond shaped line indicates
the COMPOW Protocol,” Proceedings of the European
the cluster head duration without load and the square shaped
Wireless Conference on Next Generation Wireless
line indicates the cluster head duration with load. The
Networks: Technologies, Protocols, Services and
results related to the above three graphs indicates that
Applications, pages 156-162, Florence, Italy, Feb. 25-
CMCSR outperforms TMPO.
28, 2002.
[6] H. Takagi and L. Kleinrock, “Optimal Transmission
Ranges for Randomly Distributed Packet Radio
5. Conclusions Terminals,” IEEE Transactions on Communications,
32(3),7, Mar. 1984.
The cluster size load balancing heuristics have been [7] L. Li, V. Bahl, Y.M. Wang, and R. Wattenhofer,
proposed for ad hoc networks. The cluster election heuristics “Distributed Topology Control for Power Efficient
favor the election of cluster heads based on node willingness Operation in Multihop Wireless Ad Hoc Networks,”
and number of neighbors. Here the heuristic places a cluster Proceedings of IEEE Conference on Computer
size budget on the contiguous amount of time that a node Communications (INFOCOM), Apr. 2001.
acts as cluster head. As seen from the simulation results, [8] R. Ramanathan and R. Rosales-Hain, “Topology
this heuristic produce larger cluster head durations while Control of Multihop Wireless Networks using Transmit
decreasing the cluster size and enhancing the stability. Power Adjustment,” Proceedings of IEEE Conference
on Computer Communications (INFOCOM), IEEE,
Our proposed CMCSR is a novel energy-aware topology Mar. 26-30, 2000.
management approach based on dynamic node priorities and [9] R. Prakash, “Unidirectional Links Prove Costly in
cluster size load in ad hoc networks. CMCSR consists of two Wireless Ad-Hoc Networks,” Proceedings of the
parts that implement the MDS and CDS elections Discrete Algorithms and Methods for Mobile
respectively. Compared to five prior heuristics of MDS and Computing and Communications - DialM, Seattle, WA,
CDS elections in ad hoc networks, MDS offers four key Aug. 20, 1999.
advantages. i) CMCSR obtains the MDS and CDS of the [10] S. Bandyopadhyay and E. J. Coyle, “An Energy
network without any negotiation stage; only two-hop Efficient Hierarchical Clustering Algorithm for
neighbor information is needed. ii) CMCSR allows nodes in Wireless Sensor Networks”, In Proc. INFOCOM 2003,
the network to periodically re-compute their priorities, so as San Francisco, Apr, 2003.
to balance the cluster head role and prolong the battery life [11] M. Maeda and Ed Callaway, "Cluster Tree Protocol
of each node. iii) CMCSR introduces the willingness value (ver.0.6)",http://www.ieee802.org/15/pub/2001/May01/
of a node, which decides the probability of the node being 01189r0P80215_ TG4-Cluster-Tree-Network.pdf.
elected into the MDS according to the battery life and [12] L. Bao and J.J. Garcia-Luna-Aceves, “Transmission
mobility of the node and iv) MDS introduces doorway Scheduling in Ad Hoc Networks with Directional
concept for the CDS in addition to the well-known gateway Antennas,” Proc. ACM Eighth Annual International
and cluster head concepts. Conference on Mobile Computing and networking,
A key contribution of this work consists of converting the Atlanta, Georgia, USA, Sep, 23-28 2002.
static attributes of a node, such as node identifier, into a [13] B. Chen, K. Jamieson, H. Balakrishnan, and R.
dynamic control mechanism that incorporates the three key Morris, “Span: an Energy-Efficient Coordination
factors for topology management in ad hoc networks -- the Algorithm for Topology Maintenance in Ad Hoc
nodal battery life, mobility, and cluster size load balancing. Wireless Networks,” In Proc. 7th ACM MOBICOM,
Although existing proposals have addressed all these Rome, Italy, Jul, 2001.
aspects, CMCSR constitutes a more comprehensive [14] C.C. Chiang, H.K. Wu, W. Liu, and M. Gerla,
approach. “Routing in Clustered Multihop, Mobile Wireless
Networks with Fading Channel,” IEEE Singapore
References International Conference on Networks SICON'97, pages
[1] http://w3.antd.nist.gov/wahn_mahn.shtml. 197-211, Singapore, Apr. 14-17, 1997.
[2] Tomas Johansson and Lenka Carr-Motyˇckov´. “On [15] L. Bao and J.J. Garcia-Luna-Aceves, “Topology
Clustering in Ad Hoc Networks,” First Swedish Management in Ad Hoc Networks,” Proc of the 4th
National Computer Networking Workshop, ACM Interational Symposium on Mobile Ad Hoc
SNCNW2003, 8-10 September, 2003. Networking and Computing (MOBIHOC), Annapolis,
[3] R.Pandian, P.Seethalakshmi and V.Ramachandran, Maryland, USA, Jun. 2003.
“Enhanced Routing Protocol for Video Transmission
over Mobile Adhoc Network,” Journal of Applied
64 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 4, April 2010

Authors Profile
Mr D K L V Chandra Mouly received M.Tech (CSE) from S V
University College of Engineering, Tirupati, India in the year
2007. Currently he is pursuing his Ph.D. (Part-time) at S V
University, Tirupati. His areas of interests are Computer Networks
and Distributed Systems.

Dr Ch D V Subba Rao received Ph.D (CSE) from S V University,


Tirupati, India in 2008. He got 18 years of teaching experience. At
present, he is working as Associate Professor, Dept of Computer
Science and Engineering, S V University College of Engineering,
Tirupati, India. His areas of interests are Distributed Systems,
Operating Systems, Computer Networks and Programming
Language Concepts.

Dr M M Naidu received Ph.D (IIT-Delhi) in the year 1988. He got


32 years of teaching experience. Currently he is working as
Professor in the Dept of Computer Science and Engineering, S V
University College of Engineering, Tirupati, India. His areas of
interests include Software Engineering, Enterprise Resource
Planning, Computer Networks and Computer Graphics.

Das könnte Ihnen auch gefallen