IJCSI International Journal of Computer Science Issues, Vol.

7, Issue 3, No 6, May 2010 27

ISSN (Online): 1694-0784
ISSN (Print): 1694-0814

Improving Network Performance using ACO Based

Redundant Link Avoidance Algorithm
Chandra Mohan B1, and Baskaran R2
1
Principal, John Bosco Engineering College
Chennai, Tamil Nadu, India

2
Assistant Professor, Department of CSE, Anna University
Chennai, Tamil Nadu, India

Abstract [2]. It is unambiguous from these experimental results that

In the wide spread internet, response time and pocket loss are the Average packet loss is reaches up to 17%. When
inappropriate due to network traffic, as a result the network considering response time, less than 150ms is an
efficiency becomes worst and the system provides poor Quality appreciate value for data communication, but all
of Service (QoS). An optimal routing protocol, especially continents except North America have more than 150ms.
multipath may avoid such traffic in the network. But existing
Especially in Australia, the packet loss is 0% but response
routing protocols, both single path and multi path, concentrates
only on finding the routes based on any one or some set of time is 169ms.
metrics, that not always suitable for dynamic, cloud natured
network environment. Ant Colony Optimization (ACO) based Table 1 Average Response Time and Average Packet Loss
multipath routing protocol was suggested as an alternate to this of each continent
problem by many researchers. The multipath ACO also provides
same set of link(s) for the source to destination, so that traffic Avg. Response Avg. Packet
merging again becomes a critical problem. This paper proposes Continent
Time (in ms) Loss (%)
an optimal solution to avoid the problem of traffic merging in the
network by removing redundant link in the route. The Proposed
algorithm, called Redundant Link Avoidance (RLA) algorithm, Asia 386 17%
is an ACO based multi path routing methodology, avoiding
copious link in the suggested routes of ACO multipath protocol. Australia 169 0%

Keywords: Computer Networks, Routing, QoS, Swarm Europe 205 5%

Intelligence, Ant Colony Optimization.
North America 142 8%

1. Introduction South America 138 0%

In October 1986, network had a series of congestion

collapses, the data throughput from Lawrence Berkley Lab The loss of data packets and delay in data transmission in
to University of California, sites separated by only two the computer network becomes obvious problem due to
IMP hops, dropped from 32 kbps to 40 bps, i.e., 99.88% these reasons: {a} Bottleneck problem due to the growth
packet drop [1]. Due to this incident, the traffic of networks and increased link speeds; {b} Packet arrival
management in computer network becomes critical rate exceeds the outgoing link capacity; {c} insufficient
problem in data communication. Several research efforts memory or buffer to store arriving packets, i.e., buffer of
have proposed different approaches for traffic receiving node is not sufficient to store the incoming data
management problems, each one has its own strengths and packets; {d} network Burst; {e} Demand temporarily
limitations but optimality not yet achieved. exceeds available resources at some point in the network;
{f} Same route is used for more transmission,
The following Table 1 shows the average response time overloading; and {g} traffic merging.
and the average packet loss of each continent recorded on
12th February of 2010 at 12:40 Indian Standard Time (IST) Congestion in the computer network may be controlled [3]
either by traffic management algorithms and / or by
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 28
www.IJCSI.org

optimal routing. Traffic management provide optimal algorithm finds several shortest paths that are almost
solution to solve the facts {a c} but optimal routing may equally good, between given source to destination. As the
solve all the listed facts. All the existing traffic traffic is split over several paths, better performance can
management and routing protocol concentrates to avoid be obtained in multi-path routing.
the traffic by the way of solving one or few facts in the {a
e} except {f} and {g}. In which, the facts {d} and {e} There are several research papers [5 - 13] implemented
may be controlled, only by multipath routing algorithm but variety of multi-path algorithms for various applications,
when applying multipath there are high probability that the in which Ant Colony Optimization (ACO) based multipath
network becomes congested by the facts {f} and {g}. The [13] routing provide better result than others due to its real
combination of {f} and {g}, i.e., when two or more time computation. In which, [5], [6] are few application of
sources would simultaneously try to transmit packets to multi path that provides better performance in multimedia
one destination via a single link, a high probability that the and data communication. In order to offer high video
number of packets would exceed the packets handling quality, a certain minimum end-to-end bandwidth has to
capacity of the network and lead to heavy congestion. be provided. In bandwidth-limited networks, it is difficult
to guarantee such a bandwidth with a single path. To meet
In this paper, the proposed algorithm, swarm intelligence the bandwidth requirement, a multipath approach is
based multi path routing protocol with Redundant Link proposed by Jiancong Chen et al [5]. Ka-Cheong Leung et
Avoidance (RLA) algorithm, focused to solve this al [6], explains implementation of multipath for avoiding
important facts {f}, {g} and also the other facts {a-e}. two different traffic types, connection-oriented traffic and
This routing methodology ensures traffic free routing by mixed traffic. Connection-oriented traffic consists of
avoiding copious link. Perspective of this routing traffic from TCP connections only, while connectionless
methodology is sharing load on multiple routes to obtain traffic comes from non-TCP connections. Mixed traffic is
optimal response time and at the same time avoiding composed of both connectionless and connection-oriented
packet drop due to overloading in the network. traffic. For both multimedia and data communication,
multipath avoids traffic induced by the facts {a e}
The remaining of this paper is organized in the following shown in section 1, but there are some more situations {f
manner: Survey of various related work, its classification, g} that the network becomes congested when
and existing algorithm is explained in the Section Two. implementing multi path. This paper, further continuing
The proposed work and the algorithm used in this paper with literature survey of the various multipath routing
are described in Section Three. Conclusion and further algorithms developed recently.
studies are discussed in Section Five followed by the
performance analysis of the proposed work is shown in The selection of the routing paths is major design
Section Four. consideration that has a drastic effect on the resulting
performance. Therefore, although many flow-control
2. Related Work algorithms are optimal for a given set of routing paths,
their performance can significantly differ for different sets
The basic problem of routing is, to find the path of lowest of paths. Previous studies and proposals on multi-path
cost between any two nodes. Routing algorithms can be routing in the previous context have focused on heuristic
grouped into three major classes [4]: non-adaptive, methods. In [7], a multi-path routing scheme, termed equal
adaptive and distributed. Non-adaptive algorithm, also cost multi-path (ECMP), has been proposed for balancing
static routing is computed in advance, offline using local the load along multiple shortest paths using a simple
information, their routing decision not based on round-robin distribution. By limiting itself to shortest
measurements or estimates of the current traffic and paths, ECMP considerably reduces the load balancing
topology. Adaptive algorithms, centralized routing will capabilities of multi-path routing; moreover, the equal
collect information from the entire subnet using global partition of flows along the shortest paths, resulting from
information in an attempt to make optimal solution. the round robin distribution, further limits the ability to
Distributed algorithm uses a mixture of local and global decrease congestion through load balancing.
information.
OSPF-OMP [8] allows splitting traffic among paths
Based on the discovery of no of path, the routing unevenly; however, the traffic distribution mechanism is
algorithms further divided as single-path and multi-path. based on a heuristic scheme that often results in an
Single path routing algorithm chooses only one best inefficient flow distribution. A. E. I. Widjaja [9]
shortest route between a given pair of source to destination considered multi-path routing as an optimization problem
for all of its traffic. Multi-path, the bifurcated routing with an objective function that minimizes the congestion
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 29
www.IJCSI.org

of the most utilized link in the network; however, they Initially, three different versions of AS were proposed,
focused on heuristics and did not consider the quality of namely ant density, ant quantity and ant cycle [25]. In the
the selected paths. ant density and ant quantity versions, the ants updated the
pheromone directly after a move from one city to an
Israel Cidon et al [10] concluded two phenomena in the adjacent city, in the ant cycle version the pheromone
implementation of multi path routing protocol: 1) as the update was only done after all the ants had constructed the
link capacity increases, the relative performance of multi- tours and the amount of pheromone deposited by each ant
path routing decreases, 2) as the number of possible routes was sent to be a function of the tour quality. Now a day,
increases, the relative performance of multi-path routing when referring to ACO, it means to ant cycle since the two
also increases till k number of paths and when it exceeds other variants were abandoned because of their inferior
the limit, the performance will be degraded. So to choose performance. Solution construction and Pheromone
only k path is an important consideration for implementing Update are the two phases of ACO implementation.
multi path routing and the optimal value of k, may change
in practice. Weijia Jia et al [11] proposed an approach In ACO based routing algorithm, solution construction
that named as integrated routing, to select a subset of handles formation of routing table, Pheromone update
routers to carry out multi path routing (MPR), the rest of used to choose the single optimal path based on density of
the routers execute Single path routing (SPR). With proper pheromone, in case of single path routing; and the
selection of MPR routers, shortcomings in the MPR will pheromone update desires the probability of load sharing
be rectified. This approach may provide better result but to along multiple path, in case of multi path routing. The
implement this protocol in the network requires two tasks: ACO routing may provide feasible solution for traffic
router assignment and routing interface selection. The management problem, even though the overloading {f}
performance of this approach lies only in these two tasks and traffic merging {g} still is an issue in computer
and implementing these two tasks in the internet like network. This paper concentrates to solve overloading and
environment is highly impractical. traffic merging by implementing RLA algorithm as
creamy layer over existing ACO. The following section
Dijkstra-old-touch-first (Dijkstra-OTF) with multipath describes our proposed work in detail.
extension [12] is an extended version of conventional
Dijikstras shortest path algorithm that computing all 3. Proposed Work
lexicographic- lightest paths from a source to every other
node in the network, but it requires additional The proposed routing protocol has two components,
computational efforts. DiCaro et al [13] developed Swarm namely ACO Based Multipath Routing algorithm and
Intelligence (SI) based routing algorithm, called Ant-Net redundant link avoidance algorithm. The main idea of
Routing using ant colony optimization technique with ACO is to model the problem as the search for a minimum
multi path capability. The ACO era starts with cost path in a graph. Artificial ants walk through this
Schoonderwoerd et al [14], they proposed Ant based graph, looking for good paths. Each ant has a rather
control (ABC) for telephone network, it continues with ant simple behavior so that it will typically only find rather
based algorithm for packet switched network proposed by poor-quality paths on its own. Better paths are found as
Subramanian et al [15] and a co-operative Asymmetric the emergent result of the global cooperation among ants
Forward (CAF) model for routing with asymmetric costs in the colony.
are ant colony optimization based single path routing
algorithm developed by Heusse et al [16]. The behavior of artificial ants is inspired from real ants,
they lay pheromone trails on the graph edges and choose
ACO is a stochastic approach that has been proposed to their path with respect to probabilities that depend on
solve different hard combinatorial optimization problems pheromone trails and these pheromone trails progressively
in almost all engineering fields such as traveling salesman decrease by evaporation. In addition, artificial ants have
problems, graph coloring problems, quadratic assignment some extra features that do not find their counterpart in
problems, resource-constrained project scheduling real ants. In particular, they live in a discrete world and
problem [17], knowledge representation tools[18], their moves consist of transitions from nodes to nodes.
Computer-assisted testing systems[19], Grid Workflow Also, they are usually associated with data structures that
Scheduling Problem[20], Self-Structured P2P Information contain the memory of their previous actions. In most
System[21], Intelligent transportation systems[22], Robot cases, pheromone trails are updated only after having
Path Integration[23], and Clustering and Data mining[24]. constructed a complete path and not during the walk, and
the amount of pheromone deposited is usually a function
of the quality of the path. Finally, the probability for an
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 30
www.IJCSI.org

artificial ant to choose an edge often depends not only on step is carried out until front reaches rear of the queue.
pheromones, but also on some problem-specific local The detailed algorithm is shown in the following section
heuristics. 3.1 and various test cases for the proposed work is shown
in section 3.2 to 3.4.
The ACO based multipath algorithm implemented in this
paper, using forward and backward ants. Ants in each set 3.1 Redundant Link Avoidance (RLA) Algorithm:
possess the same structure, but they are differently situated
in the environment, i.e., they can sense different inputs Procedure RLA(t, tend, t)
and they can produce different, independent outputs. Ants input t // current time
communicate in an indirect way, according to the dynamic input tend // time length of simulation
paradigm, through the information they concurrently read input t // time interval between ants generation
and write on the network nodes they visit. While travelling
toward their destination nodes, the forward ants keep // concurrent activity over the network
memory of their paths and of the traffic conditions found. for each i C do
In single path routing protocol, an optimal path recorded
by the ant was chosen as path between given source to M InitLocalTrafficModel
destination. In multi path routing protocol, all paths are T A InitNodeAvailRoutingTable
chosen for communicating from source to destination. As TS InitNodeSelectedRoutingTable
discussed in the previous section, some of the available
path in the multi path routing protocol, heavily congested // concurrent activity on each node
due to traffic merging problem.
while t tend do
in_parallel
The redundant link avoidance algorithm removes such
links called replicating links from destination towards
if (t mod t) = 0 then
source using two data structures named available route
array and selected route array. All available routes found destinationSelectDestination
by multipath routing protocol from source to destination is (traffic_distribution_at_source)
stored in the available routes, a two dimensional array,
followed by initializing a selected route (one more two Launch_Forward_Ant (source, destination)
dimensional array), and two queues namely selected_
relationship and available_ relationship. end-if

The first optimal route is copied from available route to for each (Active_Forward_Ant [source, current,
selected route, then all links of this route is divided from destination] ) do
destination towards source and stored into selected_
relationship queue and the same was removed available while (current destination) do
route. The same way, next route from available route is next_hop SelectLink (current, destination,
selected and divided. If front of available_relationship and link_queues, T A)
front of selected_relationship are equal then this route is PutAntOnLinkQueue (current, next_hop)
removed from available route, and deleted all association WaitOnDataLinkQueue(current, next_hop)
from available_relationship and the previous process is CrossLink (current, next_hop)
repeated until reaches the tail. If front of Memorize (next_hop, elapsed_time)
available_relationship and front of selected_relationship current next_hop
are not equal then next link is processed, and the above end-while

1.3

1.1 1.2 1.4 1.5

Fig 1. Data Flow in ACO based Routing

Where 1.1 Forward Ant Generation
1.2 Pheromone Matrix Formation
1.3 Removing Cyclic Path
1.4 Backward Ant Generation
1.5 Formatting Routing Table
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 31
www.IJCSI.org

LaunchBackwardAnt (destination, source,

memory_data) end-procedure

end for each 3.2 Test case 1:

for each (ActiveBackwardAnt [source, current,

destination]) do

while (current destination) do

next_hop PopMemory
WaitOnHighPriorityLinkQueue (current, next_hop)
CrossLink (current, next_hop)
from next_hop
current next_hop
UpdateTrafficModel (M,current,from,source,
memory_data) Fig 2 Data flow from source 13 to destination 5 (Test Case 1)
rGetNewPheromone (M, current, from, source, using three selected path
memory data)
UpdateLocalRoutingTable (T A, current, source, Let consider Fig 2, the following table shows the available
r) routes, directed paths of existing ACO and selected
routes, directed routes of ACO with RLA from source 13
end-while to destination 5:
end-foreach
end-in_parallel Available Route (ACO):
13 7 8 1 6 5
T S [0] = T A [0] 13 7 8 1 3 5
T SL = each link from Destination to source on T S 13 7 8 1 2 5
13 7 8 1 3 2 5
for each (T A [source, intermediate, destination]) do 13 7 8 1 3 6 5
13 7 8 1 6 3 5
T AL[] = each link from Destination to source on 13 7 8 1 2 3 5
current T A[] 13 7 8 1 6 3 2 5
Current1 First Link in T AL 13 7 8 1 2 3 6 5
Current2 First Link in T SL 13 7 12 10 8 1 6 5
13 7 12 10 8 1 3 5
for each (Current1 Null, Current2 Null) 13 7 12 10 8 1 2 5
13 7 12 10 8 1 3 2 5
if (Current1 Current2) do 13 7 12 10 8 1 3 6 5
Current1 next Link in T AL 13 7 12 10 8 1 6 3 5
Current2 next Link in T SL 13 7 12 10 8 1 2 3 5
else-if
13 7 12 10 8 1 6 3 2 5
Flag = 0
13 7 12 10 8 1 2 3 6 5
end-if
end-foreach
Selected Route (ACO with RLA):
13 7 8 1 6 5
if (Flag = 1)
T S = T A [current] 13 7 8 1 3 5
end-if 13 7 8 1 2 5

end-for each If packets are transmitted in all above 18 routes of

available routes, then the link 3 5, 2 3 and 6 5
end-while become heavy loaded and it may lead to packet drop. In
order to avoid traffic on these links, our redundant link
end-for each avoidance algorithm removes such link from routing table.
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 32
www.IJCSI.org

3.4 Test case 3:

3.3 Test case 2:
The following table 4, shows the available routes,
The following table 3, shows the available routes, directed routes of existing ACO and selected routes,
directed routes of existing ACO and selected routes, directed routes of ACO with RLA In multiple route
directed routes of ACO with RLA In multiple route routing the following are available routes from source 0 to
routing the following are available routes from source 11 destination 1:
to destination 4:
Available Route (ACO):
Available Route (ACO): 02 1
11 9 14 0 2 5 4 0 2 3 1
11 9 14 0 2 3 5 4 02561
11 9 14 0 2 1 6 5 4 02 36 1
11 9 14 0 2 1 3 5 4 02 53 1
11 9 14 0 2 3 6 5 4
11 9 14 0 2 1 6 3 5 4 Selected Route (ACO with RLA):
11 9 14 0 2 3 1 6 5 4 02 1
11 9 14 0 2 1 3 6 5 4 0 2 3 1
02561
Selected Route (ACO with RLA):
4. Result and Analysis
Response Time in Ant-Net Vs RLA on different Load condition

160
140 We implemented the proposed work in NS2 simulation, a
Response Time in ms

120
100
in Ant-Net
widely used open source discrete event Network Simulator
80
60
in RLA tool. There are many network simulation tools available,
40
20
for example, GloMoSim, Omnet, Opnet, Qualnet. In
0
100 150 200 500 1000 which NS2 is preferred by almost all researchers and
Load in Packets
academicians, because it covers a very large set of
applications, protocols, network types, network elements
(a) Ant Net Vs RLA and traffic models.
Response Time in OSPF Vs RLA on different Load condition

250
The following events are scheduled in the Implementation,
Response Time in ms

200

150
1) Data communication from Source1 to destination for 8
in OSPF

100
in RLA sec, 2) the proposed work implemented for 8 Sec. By
50
using AWK Programming, the packet loss and average
0
response time of before and after implementing the
100 150 200 500 1000
proposed algorithm is measured from the trace file. These
Load in Packets
values are listed here:
(b) OSPF Vs RLA 100 packets loss in 8 sec simulation implementing
Response Time in BF Vs RLA on different Load condition existing ACO based multi-path algorithm
300 0 packets loss in 8 sec simulation implementing
250
proposed ACO with RLA
Response Time in ms

RTT is 0.1581 Sec in existing ACO multi-path

200
in RIP (BF)
150
in RLA
100 algorithm
50
RTT is 0.1586 Sec in proposed ACO with RLA
0
100 150 200 500 1000
Load in Packets
The detailed performance analysis is shown in the
(c) BF (RIP) Vs RLA following section 4.1
Fig. 3: Response Time between various
routing protocols Vs RLA on 4.1 Simulation Result of Redundant Link Avoidance
different Load condition Algorithm Based Routing
11 9 14 0 2 5 4
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 33
www.IJCSI.org

In this section, the result of RLA algorithm from source 13 References

to destination 5, described in Test case 1 is shown Fig 2.
The data communication from source node 13 to
[1] Jacobson. V, Congestion Avoidance and Control, ACM
destination node 5 through three selected route explained
SIGCOMM Computer Communication Review, Vol 25,
in case 1, also shown in the Fig 2. Comparison of response
p157 187, 1995
time in existing ACO (Ant Net) Vs proposed ACO with
[2] Real time traffic monitoring available at
RLA algorithm is shown in the Fig. 3(a). Similarly,
http://www.internettrafficreport.com/details.htm
Comparison of response time in OSPF and SPI Vs
[3] Andrew S. Tanenbaum, Computer Networks, Prentice
proposed ACO with RLA algorithm is shown in the
Hall of India, 3rd Edition, 1998
Fig.3(b) and Fig. 3(c) respectively.
[4] Larry L. Peterson & Bruce S. Davie, Computer Networks
A Systems approach, Morgan Kaufmann Series, 2000
5. Conclusion [5] Jiancong Chen, S.-H. Gary Chan, and Victor O. K. Li,
Multipath Routing for Video Delivery Over Bandwidth-
Network traffic is a common problem in all type of Limited Networks, IEEE Journal on Selected Areas in
network, of topology, of protocol. It requires more Communications, Vol. 22, No. 10, December 2004
attention, as it will diminish all major metric of network. [6] Ka-Cheong Leung, and Victor O.K. Li, Generalized Load
Especially in internet like huge, dynamic and cloud like Sharing for Packet-Switching Networks I: Theory and
network, traffic is an unavoidable condition. This paper Packet-Based Algorithm, IEEE Transactions on Parallel
analyses the various factors that causes for congestion and and Distributed Systems, Vol. 17, No. 7, July 2006
provides optimal solution called redundant link avoidance [7] J. Moy, OSPF Version 2, IETF RFC 2328, Apr. 1998
algorithm over the existing ant colony optimization based [8] C. Villamizar, OSPF optimized multi-path (OSPF-OMP),
routing protocol. When comparing the existing routing Internet Draft, 1998.
protocol with the proposed work, the response time get [9] A. E. I. Widjaja, Mate: MPLS adaptive traffic engineering,
delayed and this delay in transmission time is a negligible, Internet Draft, Aug. 1998.
only 5 ms but all packet loss is avoided. [10] Israel Cidon, Raphael Rom, and Yuval Shavitt, Analysis of
Multi-Path Routing, IEEE/ACM Transactions On
It will perform even better, when implementing the same Networking, Vol. 7, No. 6, December 1999
work for some complex network with different topologies [11] Weijia Jia, Dong Xuan and Wei Zhao, Integrated Routing
and high packet loss. As shown in the test case 2 of Algorithms for any cast Messages, IEEE Communications
section 3.3, there are few situations, the multi route Magazine, January 2000
routing protocol may act as single route routing protocol. [12] Joao Luis Sobrinho, Algebra and Algorithms for QoS Path
It may lead to poor response time. But in the huge Computation and Hop-by-Hop Routing in the Internet,
resource centric world wide web, it is promised that it IEEE / ACM TRANSACTIONS ON NETWORKING,
never will ensue. VOL. 10, NO. 4, AUGUST 2002
[13] Di caro G., and Dorigo M. (1998), Extending AntNet for
Acknowledgement: best effort quality of service routing, First international
workshop on ant colony optimization
The First Author, Chandra Mohan B would like to thanks [14] Schoonderwoerd R., Holland O., Bruten J., (1997) Ant like
Dr. Sridharan R, Assistant Professor, Department of agents for load balancing in telecommunication networks,
Electronics and Communication Engineering, College of In proceedings of the first int. conf. on autonomous agents,
Engineering, Anna University, Chennai for his constant pp209-216, New York, ACM Press
support and help. [15] Subramanian D., Druschel P., and Chen J., (1997) Ants and
reinforcement learning: A case study in routing in dynamic
Both Authors would like to thanks Dr. Srivatsa, Professor networks, In proceedings of the 15th int. joint conf. on
(Retired), Department of Electronics and Communication artificial intelligence, pp823-838, San Francisco, Morgan
Engineering, Madras Institute of Technology, Anna Kaufmann
University, Chennai and Dr. Kannan. A, Professor, [16] Heusse M., Snyers D., Guerin S., and Kuntz P. (1998),
Department of Computer Science and Engineering, Adaptive agent driven routing and load balancing in
College of Engineering, Anna University, Chennai communication networks, Advances in complex systems,
1(2), pp 237-254.
[17] Wei-Neng Chen, Jun Zhang, Henry Shu-Hung Chung, Rui-
Zhang Huang, and Ou Liu, Optimizing Discounted Cash
Flows in Project SchedulingAn Ant Colony Optimization
Approach, IEEE TRANSACTIONS ON SYSTEMS,
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 34
www.IJCSI.org

MAN, AND CYBERNETICSPART C: APPLICATIONS about 15 years. Presently he is

AND REVIEWS, VOL. 40, NO. 1, p64 77, JANUARY serving as Principal in John
2010 Bosco Engineering College,
[18] Pedro C. Pinto, Andreas Nagele, Mathaus Dejori, Thomas
Tiruvallur, Tamilnadu, India.
A. Runkler, and Joao M. C. Sousa, Using a Local
His experience includes as
Discovery Ant Algorithm for Bayesian Network Structure
Learning, IEEE TRANSACTIONS ON Faculty in Anna University
EVOLUTIONARY COMPUTATION, VOL. 13, NO. 4, Chennai, as Tech In-charge in
p767, AUGUST 2009 Kuwait National Petroleum
[19] Xiao-Min Hu, Jun Zhang, Henry Shu-Hung Chung, Ou Liu, Company, as Principal in PTR
and Jing Xiao, An Intelligent Testing System Embedded College of Engg and Tech. His
With an Ant-Colony-Optimization-Based Test Composition present research included Networks, Data Mining and
Method, IEEE TRANSACTIONS ON SYSTEMS, MAN, Swarm intelligence. He has published dozen of papers in
AND CYBERNETICSPART C: APPLICATIONS AND National, International Conferences and Book series. His
REVIEWS, VOL. 39, NO. 6, NOVEMBER 2009
paper is awarded as best paper in the IEEE International
[20] Wei-Neng Chen, and Jun Zhang, An Ant Colony
Optimization Approach to a Grid Workflow Scheduling
Conference (SNPD 2008). He is a reviewer in IEEE
Problem With Various QoS Requirements, IEEE JSAC, Elsevier and few international journals. He
TRANSACTIONS ON SYSTEMS, MAN, AND delivered more than 50 Special Lectures in National,
CYBERNETICSPART C: APPLICATIONS AND International Seminars, Workshops and Faculty
REVIEWS, VOL. 39, NO. 1, JANUARY 2009 Development Programs. He has worked in Kuwait. He is a
[21] Agostino Forestiero, and Carlo Mastroianni, A Swarm life member of Institution of Electronics and
Algorithm for a Self-Structured P2P Information System, Telecommunication Engineers (IETE), Indian Society for
IEEE TRANSACTIONS ON EVOLUTIONARY Technical Education (ISTE), Computer Society of India
COMPUTATION, VOL. 13, NO. 4, AUGUST 2009
(CSI) and International Association for Engineers
[22] Minglu Li, Hongzi Zhu, Member, Yanmin Zhu, and Lionel
M. Ni, ANTS: Efficient Vehicle Locating Based on Ant
(IAEng).
Search in Shanghai Grid, IEEE TRANSACTIONS ON
VEHICULAR TECHNOLOGY, VOL. 58, NO. 8, [**]Dr. R. Baskaran received
OCTOBER 2009 his B.Tech in Electrical and
[23] Girma S. Tewolde, and Weihua Sheng, Robot Path Electronics Engineering,
Integration in Manufacturing Processes: Genetic Algorithm Master Degree in Computer
Versus Ant Colony Optimization, IEEE TRANSACTIONS Science and Engineering,
ON SYSTEMS, MAN, AND CYBERNETICSPART A: Madras University and
SYSTEMS AND HUMANS, VOL. 38, NO. 2, MARCH
Doctorate from Anna
2008
University in the years 2000,
[24] Parag M. Kanade and Lawrence O. Hall, Fuzzy Ants and
Clustering, IEEE TRANSACTIONS ON SYSTEMS, 2001, and 2007 respectively.
MAN, AND CYBERNETICSPART A: SYSTEMS AND He is now working as an
HUMANS, VOL. 37, NO. 5, SEPTEMBER 2007 Assistant Professor in Department of Computer Science
[25] Dorigo M and Stutzle T, Ant Colony Optimization, MIT and Engineering, College of Engineering Guindy, Anna
Press, First Edition University Chennai. His present research includes
Networks, Database, Data mining and warehousing and
Authors Image Processing. He presented more than 50 Special
Lectures in National, International Seminars, Workshops
[*] Mr. B. Chandra Mohan received his Diploma in and Development Programs. He is an expert in Data
Electrical and Electronics Engineering, B.Tech Degree in mining. He has published 35+ papers in International,
Information Technology and Master degree in Computer National Journals and Conferences. He is a reviewer in
Science and Engineering from Directorate of Technical IEEE/ ACM JSAC, Elsevier and many international
Education, Madras University and Anna University in the journals. He is a life member of Institution of Electronics
years 1993, 2003 and 2006 respectively. Presently he is and Telecommunication Engineers (IETE), Indian Society
pursuing Research in Anna University Chennai. He for Technical Education (ISTE), Computer Society of
worked both industry and academic in India and Abroad
IJCSI International Journal of Computer Science Issues, Vol. 7, Issue 3, No 6, May 2010 35
www.IJCSI.org

India (CSI) and International Association for Engineers

(IAEng).