Vous êtes sur la page 1sur 5

2012 International Conference on Computing Sciences

Performance Evaluation Of AntHocNet Routing Algorithm in Ad Hoc Networks

T.Nishitha, P.Chenna Reddy


Computer Science and Engineering
JNTUACEP
Pulivendula, India
e-mail: naishi.123@gmail.com, pcreddy1@rediffmail.com

Abstract—Ad hoc wireless multi-hop networks (AHWMNs) coordinator that manage routing decisions. AHWMN routing
are communication networks that consist of wireless nodes, protocols are classified as topology-based, position-based
placed together in an ad hoc manner, i.e. with minimal prior
and bio-inspired routing protocols.
planning. All nodes in the network have routing capabilities and
forward data packets for other nodes in multi-hop fashion.
AHWMNs pose substantially different challenges to routing
protocols than more traditional wired networks. AHWMN
routing protocols are classified as topology-based, position-based
and bio-inspired routing protocols. Topology-based routing
protocols use the information about the links that exist in the
network to perform packet forwarding. Position-based routing
algorithms require information about the physical position of the
participating node. Bio-inspired networks are one engineering
field which has many parallels with biology and hence the
solutions of biology can be used to solve the problems of
computer networks.
Ant Colony Optimization (ACO) is a metaheuristic inspired by
the shortest path searching behaviour of various ant species. Ant
Colony Optimization (ACO) routing algorithm is one of the
adaptive and efficient routing algorithms for AHWMNs. In ACO Figure 1. Classification of routing protocols.
routing algorithms, ant-like agents traverse the network to search
a path from source to destination, and lay down pheromone on Topology-based routing protocols use the information
the path. A data packet is transferred along a path selected with about the links that exist in the network to perform packet
probability based on the amount of pheromone. forwarding. They can be further divided into proactive,
AntHocNet is based on Ant Colony Optimization technique. It is reactive and hybrid approaches.
a hybrid algorithm that combines a reactive route setup process Proactive algorithms employ classical routing strategies
with a proactive route maintenance process. The reactive route such as distance- vector routing (e.g. DSDV) or link-state
setup is carried out at the start of a communication session or routing (e.g. OLSR). They maintain routing information
whenever the source of a current session has no more routing about the available paths in the network even if these paths
information available for the destination. The proactive route are currently not used. The main drawback of these
maintenance is run for the entire duration of the session. Its aim approaches is the maintenance of unusual path may occupy a
is to keep information about existing routes up to date and significant part of the available bandwidth if the topology of
explore new routes. In this paper performance of AntHocNet the network changes frequently.
with the Ad hoc On demand distance Vector (AODV) routing Reactive routing protocols such as AODV and DSR
protocol and Dynamic Source Routing (DSR) protocol is maintain only the routes that are currently in use and hence
compared.
reduce the burden on the network. However, they still have
some inherent limitations. First, since the routes are
Keywords- maintained only while in use, it is required to perform a route
Ant,AHWMN,AntHocNet,Performance,AODV,DSR. discovery before packets are exchanged between
communication nodes. Second, even though route discovery
is restricted to the routes currently in use, it may still
I. INTRODUCTION generate a significant amount of network traffic when the
An ad hoc wireless multi-hop network (AHWMNs) is a topology of the network changes frequently.
collection of mobile devices which form a communication Hybrid ad hoc routing protocols such as ZRP combine
network with no pre-existing wiring or infrastructure. local proactive routing and global reactive routing in order to
Routing in AHWMNs is challenging since there is no central achieve a higher level of efficiency and scalability. However,

978-0-7695-4817-3/12 $26.00 © 2012 IEEE 207


DOI 10.1109/ICCS.2012.58
even a combination of both strategies still needs to maintain ARA works in an on demand way. It sets up multiple
at least those network paths that are currently in use, limiting paths between source and destination. At the start of the data
the amount of topological changes that can be tolerated session, FANTS (Forward ants) will be broadcasted by
within a given amount of time. sender to all its neighbors. Each FANT has a unique
Position-based routing algorithms eliminate some of the sequence number to avoid duplicates. A node receiving a
limitations of topology-based routing by using additional FANT for the first time creates a record consists of
information. They require information about the physical destination address, next hop, pheromone value in its routing
position of participating nodes. table. The node interprets the source address of the FANT as
Routing of packets in AHWMNs with a large number of destination address, the address of the previous node as next
nodes or with high mobility is a very difficult task and hop, and computes the pheromone value depending on the
current routing protocols do not really scale well with these number of hops the FANT needed to reach the node. When
scenarios. Bio-inspired routing protocols solve this problem. the FANT reaches destination, the destination node extracts
These protocols can solve problems like Traveling Salesmen the information and then destroys the FANT. A BANT
Problem, the vehicular routing problem, the graph colorings (Backward ANT) is created and sent towards the source
problem, the shortest common super-sequence problem, the node, the path is established and data packets are sent. Data
multiple knapsack problems, the bin packing problem, the packets are used to maintain the path, so overhead is
2D HP protein folding problem, etc... reduced.
The Ant colony optimization (ACO) is based on the AntNet is an Ant Colony Optimization (ACO)
foraging behavior of ants. When ants search for food, they metaheuristic for data network routing proposed by Gianni
wander randomly and upon finding food return to their Di Caro and Marco Dorigo. It was developed for packet
colony while laying a chemical substance called pheromone. switched wired networks. In this, at regular intervals of time,
Many ants may travel through different routes to the same from every network nose s, a forward ant fs->d is sent
food source. The ants, which travel the shortest path, toward a destination d to discover a feasible, low-cost path to
reinforce the path with more pheromone that aids other ants that node, and to investigate the load status of the network
to follow. Subsequently more ants are attracted by this along the path. While traveling towards their destination
pheromone trail, which reinforces the path even more. This nodes, the forward ants keep memory of their paths and of
autocatalytic behavior quickly identifies the shortest path. the traffic conditions found. When an ant reaches a node that
Ants are simple autonomous agents that interact via indirect is already in memory, a cycle is detected and all the nodes
communication known as stigmergy. Stigmergy is an until this recurrent node are deleted from the ant’s memory.
indirect form of communication where individual agents When the destination node d is reached, the agent Fs->d
leave signals in the environment and other agents sense them generates a backward ant Bd->s, transfers to it all its
to drive their own behavior. This form of communication is memory, and is deleted. The backward ant takes the same
local wherein simple agents interact locally without having path as that of its corresponding forward ant, but in the
any global information. opposite direction. The backward ant updates the traffic and
Since these bio-inspired networks are scalable, it will be the pheromone value for all the entries corresponding to the
helpful for further research also. destination node d.
AntNet–FA is an adoption of AntNet, proposed by the
II. LITERATURE REVIEW same authors of original algorithm. It is very similar to
The robustness and efficiency of the collective behaviors original AntNet, but contains an improvement in behavior of
of insect societies with respect to variations of environment forward ants: AntNet-FA's forward ants do not use the same
conditions is a key aspect of their biological success. queues as data packets, but instead take high priority queues
Because of these same properties, they have recently become like backward ants. The trip times experienced by the
a source of inspiration for the design of routing algorithms forward ants are therefore no longer representative for what
for dynamic networks. Ant algorithms are differed according can be expected for data packets; the trip time for data
to how pheromone is updated, how routing table packets are instead calculated by the backward ants as the
probabilities are calculated, how often and how many ants sum of local estimates maintained in each of the intermediate
sent for request. A number of different ACO algorithms for nodes. The main advantage of this approach is that ants can
adaptive routing have been proposed. travel faster so that updates are done more in real-time.
The first of these were Ant-based control (ABC). ABC Routing By Ants (RBA) uses virtual circuits and supports
Routing was developed for wired telecommunication both unicast and multicast routing. RBA has similarities both
networks and it assumes symmetric path costs between with ABC and AntNet. An interesting difference with these
nodes. Ants move from node to node, selecting the next node two algorithms is the fact that the parameters which define
to move to according to the probabilities in the pheromone how routing decisions are derived from pheromone values
tables for their destination node. When an ant arrives at a are carried inside the ants, so that they can be different for
node, the entry in the pheromone table corresponding to the each ant. These parameters are assigned to ants in their
node from which the ant has just come is increased. When source node and are calculated using a genetic algorithm
ants have reached their destination, they die. The ants get (GA).
delayed on parts of the system that are heavily used. Some Probabilistic Emergent Routing Algorithm (PERA)
noise can be added to avoid freezing of pheromone trails. algorithm works in an on-demand way, with ants being

208
broadcast towards the destination at the start of a data DSR at data rates 1Mb,10Mb,100Mb and 1000Mb in UDP
session. Multiple paths are set up, but only the one with the environment
highest pheromone value is used by data and the other paths
are available for backup. The route discovery and
maintenance is done by flooding the network with ants. Both
forward and backward ants are used to fill the routing tables
with probabilities. By using these probabilities, a neighbor
will forward a packet to the given destination. Neighbors are
discovered using HELLO messages, but entries will be
inserted in the routing table are after receiving a backward
ant from the destination node.
AntHocNet is a hybrid algorithm, containing both
reactive and proactive elements. The algorithm is reactive in
the sense that it only gathers routing information about
destinations that are involved in communication sessions. It
is proactive in the sense that it tries to maintain and improve
information about existing paths while the communication
session is going on (unlike purely reactive algorithms, which
do not search for routing information until the currently
known routes are no longer valid). Routing information is Figure 2. Packet Delivery Ratio at 1 Mb data rate
stored in pheromone tables that are similar to the ones used
in other ACO routing algorithms. Forwarding of control and
data packets is done in a stochastic way, using these tables.
Link failures are dealt with using specific reactive
mechanisms, such as local route repair and the use of
warning message.

III. SIMULATION

A. Methodology
To test and compare the performance of AntHocNet
protocol, the network simulator NS-2, version 2.34 is used.
The network model used in simulation is composed by
mobile nodes and links that are considered wireless. Each
node considered as communication end-point is host and a
forwarding unit is router.
In addition to NS-2, a set of tools, mainly Bash scripts Figure 3. Packet Delivery Ratio at 10 Mb data rate
and AWK filters, to post-process the output trace files
generated by the simulator are developed. In order to
evaluate the performance, multiple experiments have been
set up.
B. Metrics
The metrics that are used for comparing the AntHocNet,
AODV, DSR routing protocols are Packet Delivery Ratio
and End-to-End Packet Delay.

1) Packet Delivery Ratio: This is the ratio of total no of


packets successfully received by the destination nodes to
number of packets sent by the source nodes throughout the
simulation.
2) End-to-End Packet Delay: It is the cumulative
statistical measure of the delays experienced by packets
travelling between source and destination.
C. Simulation Results Figure 4. Packet Delivery Ratio at 100 Mb data rate
The following results shows the Packet Delivery Ratio
and End-to-End Packet Delay of AntHocNet, AODV and

209
Figure 5. Packet Delivery Ratio at 1000 Mb data rate Figure 8. End-to-End Packet Delay at 100 Mb data rate

Figure 6. End-to-End Packet Delay at 1 Mb data rate


Figure 9. End-to-End Packet Delay at 1000 Mb data rate

The following table shows the packet delivery ratio of


AntHocNet, AODV and DSR in static environment with six
nodes for TCP at window sizes 8, 16, 32.

Figure 7. End-to-End Packet Delay at 10 Mb data rate

210
TABLE I. PACKET DELIVERY RATIO IN TCP AT WINDOW SIZES IV. CONCLUSIONS
8,16,32.
Simulation results demonstrate the scalability of
TCP AntHocNet when compared to AODV and DSR i.e.,
AntHocNet AntHocNet performs better at high data rates and at large
Window Size Sent Received Packet Delivery
Ratio
number of nodes. Its performance is inferior to that of
8 3887 3886 0.9997 AODV and DSR at low data rates and at less number of
16 3897 3887 0.9974 nodes. From this it can be concluded that AntHocNet is
32 3910 3888 0.9944 suggested for large-scale, high data rate networks. As the
AODV
8 3893 3891 0.9995
number of nodes increases, the performance of AODV and
16 3900 3890 0.9974 DSR deteriorates whereas AntHocNet’s performance
32 3904 3894 0.9974 increases even with the increase in the number of nodes. At
DSR high data rates also AODV and DSR has very low
8 3787 3786 0.9997 performance and it deteriorates rapidly. But for AntHocNet,
16 3797 3787 0.9937
32 3806 3785 0.9944
its performance is either constant or increases at high data
rates.
REFERENCES
TABLE II. PACKET DELIVERY RATIO IN UDP AT DATA RATES 1
MB,10 MB, 100 MB [1] Frederic Ducatelle, (2007) “Adaptive Routing in Wireless Ad Hoc
Multi-Hop Networks”.
[2] R.M. Sharma, “Performance Comparison of AODV, DSR and
UDP
AntHocNet Protocols”, Department of
AntHocNet
Data Rate Sent Received Packet Delivery
Computer Engineering, NIT Kurukshetra
Ratio [3] FJ Arbona Benat, (2006) “Simulation of Ant Routing Protocol for
1 Mb 12501 3339 0.2671 Ad-hoc networks in NS-2”
10 Mb 125001 3237 0.02589 [4] Azzedine Boukerche, Begumhan Turgut, “Routing
100 Mb 1250001 3236 0.002589 protocols in ad hoc networks: A survey”
AODV
1 Mb 12501 3345 0.2676 [5] S. L. Ho, Shiyou Yang, H. C. Wong. “An Improved Ant Colony
10 Mb 125001 3345 0.02676 Optimization Algorithm and Its Application to Electromagnetic
100 Mb 1250001 3345 0.002676 Devices Designs”. Hong Kong Polytechnic University. March 14,
DSR
2009.
1 Mb 12501 3343 0.2674 [6] Christian Blum, “Ant colony optimization Introduction and recent
10 Mb 125001 3342 0.02673 trends” Universitat Politècnica deCatalunya, Spain, October 11 2005.
100 Mb 1250001 3343 0.002674

211

Vous aimerez peut-être aussi