22 IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.

2, February 2009

Performance Analysis of MANET Routing Protocols in

Different Mobility Models
M.K.Jeya Kumar † and R.S.Rajesh††
†
Research Scholar, Dr.M.G.R. University, Chennai, India.
††
Faculty of Computer Science & Engineering, Manonmaniam Sundaranar University,
Tirunelveli, India.

Summary changes the network topology drastically and unpredict-

Ad Hoc Networks are multi - hop wireless networks with ably.
dynamically changing network connectivity owing to mobility. The various ad hoc routing protocols have their unique
The protocol suite includes several routing protocols specifi- characteristics. Hence, in order to find out the most
cally designed for ad-hoc routing. The conventional routing adaptive and efficient routing protocol for the highly
protocols such as shortest-path routing algorithms are not
particularly well suited for operation in ad-hoc networking
dynamic topology in ad hoc networks, the routing
environment. The most widely used ad hoc routing pro- protocols behavior has to be analyzed using varying
tocols are Ad-hoc On-Demand Distance Vector Routing node mobility speed, Traffic and network size. Thus,
(AODV), Destination Sequenced Distance Vector (DSDV), the goal is to carry out a systematic performance com-
Dynamic Source Routing (DSR) and Temporally - Ordered parison of ad hoc routing protocols under mobility
Routing Algorithm (TORA). In this paper, the three random models.
based mobility models such as Random waypoint, Random walk The main aim of this paper is :
and Random Directions were implemented. The two different • Acquiring the detailed understanding of ad hoc
parameter constraints like packet-delivery fraction and End-to-
routing protocols
end packet delivery delay are compared with respect to mobility
speed, Traffic and Network size. The simulation results shows • Implementing the Mobility models
that the AODV protocols in Random Waypoint mobility model • Analyzing the performance differentials of
performs better than DSDV, TORA and DSR in Random walk routing protocols under mobility.
and random Direction mobility model. . Based on the observa- The organization of the paper is as follows. Section 2
tions, it is to suggest that AODV routing protocol can be discusses the major mobile Ad hoc routing protocols used
used under high mobility since it outperforms DSDV, TORA in this evaluation study. Section 3 presents the Random
and DSR protocols. mobility models used in this analysis. The simulation re-
Key words : sults, followed by their interpretations are presented in
Ad Hoc Networks, Mobility Models, AODV, DSDV, TORA
section 4. The results obtained in this simulation are also
discussed in section 5. Based on the analysis, Section 6
1 Introduction presents the conclusions.

A mobile ad hoc network (MANET) is an autonomous 2. Mobile Ad-hoc Networking Protocols

system of mobile hosts connected by wireless links.
There is no static infrastructure such as base stations. The main problem with ad-hoc networking is how to send
Each node in the network also acts as a router, for- a message from one node to another with no direct link.
warding data packets for other nodes. Any number of The nodes in the network are moving around unpredicta-
people could conceivably enter a conference room and bly, and it is very challenging which nodes that are directly
agree to support communication links between them- linked together.. The topology of an ad-hoc network is
selves, without necessarily engaging the services of constantly changing and it is very difficult for routing
any pre-existing equipment in the room. Thus, it is a process. There are two main approaches for routing
temporary network with no wires and no administra- process in ad hoc networks. The first approach is a pro-
tion intervention required. active approach which is table driven and uses periodic
A central challenge in the design of ad hoc networks protocols. This means that all nodes have tables with
is the development of dynamic routing protocols that routing information which are updated at intervals. The
can efficiently find routes between two communicating second approach is re-active, source-initiated or on-
nodes. The routing protocols must be able to cope up demand. This means that every time a message is sent it
with the high degree of node mobility that often first has to find a path by searching the entire network.
There are many different protocols that are in accordance

Manuscript received February 5, 2009

Manuscript revised February 20, 2009
IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009 23

with the two different routing approaches. Different proto- 2.2 Ad-Hoc On Demand Distance Vector Routing
cols are specialized in different aspects of the routing. – AODV
Other aspects than finding a short path are low overhead
communication and load-balancing. The Ad-hoc On-demand Distance Vector routing proto-
The AODV, TORA and DSR are source-initiated or on- col [7][1] enables multihop routing between the partic-
demand routing protocols and DSDV is a table driven ipating mobile nodes wishing to establish and maintain
protocol. The ad hoc routing protocols considered in this an ad-hoc network. AODV is a reactive protocol
study are explained below. based upon the distance vector algorithm.
The algorithm uses different messages to discover and
2.1 Destination - Sequenced Distance Vector – maintain links. Whenever a node wants to try and
DSDV find a route to another node it broadcasts a Route Re-
quest (RREQ) to all it’s neighbors. The RREQ propa-
DSDV [2] belongs to the class of pro-active routing gates through the network until it reaches the destina-
protocols. This protocol is based on the classical Bell- tion or the node with a fresh enough route to the des-
man-Ford routing algorithm [2] to apply to mobile ad tination. Then the route is made available by uncasing
hoc networks. DSDV also has the feature of the dis- a RREP back to the source.
tance-vector protocol [3] in that each node holds a The algorithm uses hello messages (a special RREP)
routing table including the next-hop information for that are broadcasted periodically to the immediate
each possible destination. Each entry has a sequence neighbors. These hello messages are local advertise-
number. If a new entry is obtained, the protocol prefers ments for the continued presence of the node, and
to select the entry having the largest sequence number. neighbors using routes through the broadcasting node
If their sequence number is the same, the protocol will continue to mark the routes as valid. If hello
selects the metric with the lowest value. messages stop coming from a particular node, the
Routing information is transmitted by broadcast. Up- neighbor can assume that the node has moved away
dates have to be transmitted periodically or immediate- and mark that link to the node as broken and notify
ly when any significant topology change is available. the affected set of nodes by sending a link failure noti-
Sequence numbers are assigned by destination, means fication (a special RREP) to that set of nodes.
the destination gives a sort of default even sequence
number, and the emitter has to send out the next up- 2.3 Temporally - Ordered Routing Algorithm –
date with this number. TORA
Packets are transmitted between the stations of the
network by using routing tables which are stored at TORA protocol [10] belongs to the class of reactive proto-
each station of the network. Each routing table, at each cols. The protocol is highly adaptive, efficient and it is
of the stations, lists all available destinations, and the used to establish the “temporal order” of topological
number of hops to each. Each route table entry is change events which is used to structure the reaction
tagged with a sequence number which is originated by to topological changes. The protocol is designed to mi-
the destination station. To maintain the consistency of nimize reaction to topological changes. The protocol is
routing tables in a dynamically topology, each station distributed in that nodes need only maintain informa-
periodically transmits updates, and transmits updates tion about adjacent nodes. The protocol is “source in-
immediately when significant new information is avail- itiated” and quickly creates a set of routes to a given
able. destination only when desired.
Routing information is advertised by broadcasting or The protocol accomplishes three functions through the
multicasting the packets which are transmitted periodi- use of three distinct control packets[8] such as query
cally and incrementally as topological changes are (QRY), update (UPD) and clear (CLR). QRY packets
detected - for instance, when stations move within the are used for both creating and maintaining routes, and
network. Data is also kept about the length of time CLR packets are used for erasing routes.
between arrival of the first and the arrival of best
route for each destination. Based on this data, a deci- 2.4 Dynamic Source Routing-DSR
sion may be made to delay advertising routes which
are about to change soon, thus damping fluctuations of Dynamic Source Routing(DSR) [5], belongs to the class of
the route tables. reactive protocols and allows to dynamically discover a
route across multiple network hops to any destination.
Source routing means that each packet in its header carries
the complete ordered list of nodes through which the pack-
et must pass. DSR uses no periodic routing of messages.,
there by reducing network bandwidth overhead, conserv-
24 IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009

ing battery power and avoiding large routing updates calculated. The movement trace of a mobile node using the
throughout the ad-hoc network. Instead DSR relies on Random Walk model is shown in figure 2.
support from the MAC layer.
1000

3 Random Mobility Model 800

The mobility model[8] plays a very important role in de- 600

termining the protocol performance in mobile Ad Hoc 400

Network. Hence, this work is done using the random mo-
200
bility models like Random Waypoint, Random Walk and
Random Direction. These models with various parameters 0
reflect the realistic traveling pattern of the mobile nodes. 0 100 200 300 400 500 600 700 800 900 1000

The following are the three models with the traveling pat-
tern of the mobile nodes during the simulation time.
Figure 1 Node Movement in Random Way Point.

3.1 Random Waypoint

1000
The Random Way Point Mobility Model includes pauses 800
between changes in direction and/or speed. A Mobile node 600
begins by staying in one location for a certain period of 400
time (i.e. pause). Once this time expires, the mobile node
200
chooses a random destination in the simulation area and a
0
speed that is uniformly distributed between [min-speed,
0 200 400 600 800 1000
max-speed]. The mobile node then travels toward the new-
ly chosen destination at the selected speed. Upon arrival,
the mobile node pauses for a specified period of time start-
ing the process again. The random waypoint model is a Figure 2 Node Movement in Random Walk
commonly used mobility model in the simulation of ad
Since many entities in nature move in extremely unpre-
hoc networks. It is known that the spatial distribution of
dictable ways, the Random Walk Mobility Model was
network nodes moving according to this model is non uni-
developed to mimic this erratic movement. An MN moves
form. However, a closed-form expression of this distribu-
from its current location to anew location by randomly
tion and an in-depth investigation is still missing. This fact
choosing a direction and speed in which to travel. The new
impairs the accuracy of the current simulation methodolo-
speed and direction are both chosen from pre-defined
gy of ad hoc networks and makes it impossible to relate
ranges, [speedmin, speedmax] and [0, 2*pi] respectively.
simulation-based performance results to corresponding
Each movement in the Random Walk Mobility Model
analytical results. To overcome these problems, it is pre-
occurs in either a constant time interval ‘t’ or a constant
sented a detailed analytical study of the spatial node distri-
distance traveled ‘d’, at the end of which a new direction
bution generated by random waypoint mobility. The
and speed are calculated. If an MN which moves accord-
movement trace of a mobile node using the Random Way-
ing to this model reaches a simulation boundary, it
point model is shown in figure 1. It is considered that a
bounces off the simulation border with an angle deter-
generalization of the model in which the pause time of the
mined by the incoming direction. The MN then continues
mobile nodes is chosen arbitrarily in each waypoint and a
along this new path. random walk on a one or two-
fraction of nodes may remain static for the entire simula-
dimensional surface returns to the origin with complete
tion time.
certainty, i.e., a probability of 1.0. This characteristic en-
sures that the random walk represents a mobility model
3.2 Random Walk that tests the movements of entities around their starting
In this mobility model, a mobile node moves from its cur- points, without worry of the entities wandering away never
rent location to a new location by randomly choosing a to return. Random Walk is a memory-less mobility pattern.
direction and speed in which to travel. The new speed and This characteristic can generate unrealistic movements
direction are both chosen from pre-defined ranges, [min- such as sudden stops and sharp turns.
speed, max-speed] and [0, 2*pi] respectively. Each move-
ment in the Random Walk Mobility Model occurs in either 3.3 Random Direction
a constant time interval ‘t’ or a constant traveled ‘d’ dis-
A mobile node chooses a random direction in which to
tance, at the end of which a new direction and speed are
travel similar to the Random Walk Mobility Model. The
IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009 25

node then travels to the border of the simulation area in 4.1 Speed vs Packet Delivery Fraction
that direction. Once the simulation boundary is reached,
the node pauses for a specified time, chooses another an- The Performance of the routing protocols in terms of
gular direction (between 0 and 180 degrees) and continues packet delivery ratio is examined with respect to the mo-
the process. bility of nodes. Tow different network traffic density
The Random Direction Mobility Model was created to scenarios are considered one with 10 connections and
overcome clustering of nodes in one part of the simulation another with 20 connections. The simulation results are
area. produced by the Random Waypoint Mobility Model. shown in the figure 4.
In the case of the Random Waypoint Mobility Model, this In Random Way point model, packet delivery ratios pro-
clustering occurs near the center of the simulation area. In duced by all the protocols are very close when the speed is
the Random Waypoint Mobility Model, the probability of low. The slight difference in the ratio is produced for with
an MN choosing a new destination that is located in the 10 connections and 20 connections. When the speed is
center of the simulation area, or a destination which re- increased to 20 m/s. the packet delivery ratio s produced
quires travel through the middle of the simulation area, is by the protocols differ sharply and this difference becomes
high. In this model, MNs choose a random direction in more with 20 connections.In the case of Random walk and
which to travel similar to the Random Walk Mobility and Random Direction mobility models, the packet deli-
Model. An MN then travels to the border of the simulation very ratio differ heavily for lower mobility and higher mo-
area in that direction. Once the simulation boundary is bility.
reached, the MN pauses for a specified time, chooses
another angular direction [0, 180] and continues the 4.2 Traffic vs Packet delivery fraction
process. In a slightly modified version MNs continue to
The performance of the routing protocols in terms of pack-
choose random directions but they are no longer forced to
et delivery ratio is examined with respect to traffic load.
travel to the simulation boundary before stopping to
Tow different network traffic density scenarios are consi-
change direction. Instead, an MN chooses a random direc-
dered one with 10 connections and another with 20 con-
tion and selects a destination any where along that direc-
nections. The simulation results are shown in the figure 5.
tion of travel. The movement trace of a mobile node using
The packet delivery ratos obtained from the simulation sho
the Random Direction model is shown in figure 3.
sharp decrease when the number of packets is increased
from 1 to 4 and number of connections is increased form
600
10 to 20. The differences in packet delivery ratios pro-
500 duced by the routing protocols are very less in Random
400 Waypoint mobility model. Larger differences in packet
300 delivery ratio are obtained in Random walk and random
200
Direction mobility models.
100
100
0
0 50 100 150 200 250 300
Avg. Packet Delivery Ratio

AODV 10
90
AODV 20
DSDV 10
DSDV 20
Figure 3 Node Movement in Random Direction 80 TORA 10
TORA 20
DSR 10
DSR 20
70
4.Performance Results
60
This section discusses the various predominance metrics 10 20
used and the Performance differentials analyzed. The per- Speed in m /s

formance metrics analyzed are the fraction of packets deli-

vered at the destination and the packet delivery ratio for
various speeds of mobility, Traffic and Network Size. a) Random way point
The simulation is done with different nodes in wireless
sensor networks with respect to the random-based mobility
model: Random Waypoint, Random Walk and Random
direction models. The protocols considered for analysis are
AODV, DSDV, TORA and DSR.
26 IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009

100
100
95 90

Avg. Packet Delivery Ratio

Avg. Pkt. Delivery Ratio

AODV 10
90 80
AODV 20 AODV 10
85 AODV 20
DSDV 10 70
80 DSDV 10
DSDV 20 DSDV 20
75 TORA 10 60 TORA 10
70 TORA 20 TORA 20
50 DSR 10
65 DSR 10
DSR 20
DSR 20 40
60
55 30
50 20
10 20
1 4
Speed in m /s Packets per second

c) Random Direction
b) Random Walk
Figure 5 Packet Delivery Fraction for varying number of sources

100
Avg. Pkt. Delivery Ratio

AODV 10 4.3 Node density Vs Packet Delivery Fraction

90 AODV 20
DSDV 10
80 DSDV 20 The performance of the Routing protocols in terms of
TORA 10
70 TORA 20
packet delivery ratio is examined with respect to the area
DSR 10 in which the nodes are likely to move. Packet delivery
60 DSR 20
ratios are considered for 10 connections and 20 connec-
50
tions traffic density. The simulation results are shown in
10 20
Speed in m/s the figure 6.
In this a higher packet delivery ratio for higher density of
nodes and decreases when the when the node density be-
c) Random Direction comes sparse. In Random waypoint mobility model
Figure 4. Packet Delivery Fraction for varying speeds
AODV produces higher packet delivery ratio and DSDV,
TORA, and DSR produces lower packet delivery ratio.
Avg. Packet Delivery Ratio

100
90 100
Avg. Pkt. Delivery Ratio

AODV 10
80 AODV 20 95 AODV 10
70 DSDV 10 90 AODV 20
DSDV 20 85 DSDV 10
60 TORA 10
80 DSDV 20
50 TORA 20 TORA 10
DSR 10 75
40 TORA 20
DSR 20 70 DSR 10
30
65 DSR 20
20
60
1 4
400 x 400 1000 x 600
Packets per second Simulation Area

a) Random way point a) Random way point

A vg. Packet D elivery Ratio

100
100
90 AODV 10 90 AODV 10
Avg Pkt Delivery Ratio

80 AODV 20 80 AODV 20
DSDV 10 DSDV 10
70
DSDV 20 70 DSDV 20
60 TORA 10 TORA 10
60
50 TORA 20 TORA 20
40 DSR 10 50 DSR 10
30 DSR 20 40 DSR 20

20 30
1 4 400 x 400 1000 x 600
Simulation Area
Packets per second

b) Random Walk
b) Random Walk
IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009 27

100

90 2700
AODV 10
Avg Pkt Delivery Ratio

AODV 20 2400
80 AODV 10
DSDV 10 2100

Avg. Delay in m/s

AODV 20
70 DSDV 20 1800 DSDV 10
TORA 10
60 1500 DSDV 20
TORA 20
1200 TORA 10
50 DSR 10
TORA 20
DSR 20 900
40 DSR 10
600
30 DSR 20
300
400 x 400 1000 x 600
0
Simulation Area
10 20
Speed in m/s
c) Random Direction
Figure 6. Packet Delivery Fraction for Varying Network Size c) Random Direction
Figure 7. End-to-end delay for varying speeds
4.4 Speed vs End-to-End Delay
4.5 Traffic vs End-to-End Delay
The performance of the routing protocols in terms of End-
to-End Delay is examined with respect to mobility of the The performance of the routing protocols in terms of End-
nodes. End-to-end delay are considered for 10 connections to-End Delay is examined with respect to traffic load.
and 20 connections traffic density. The results are shown End-to-end delay are considered for 10 connections and 20
in the figure 7. connections traffic density scenarios. The simulation re-
With Random waypoint and Random direction mobility sults are shown in the figure 8.
models all the The protocols in random waypoint takes In all mobility models the routing protocols consume less
less time to deliver the packets compared to Random walk time to deliver packets with 10 connections and 1 packets
and Random Direction mobility model. The difference in per second/connections protocols. More time is spend to
time used by DSDV, TORA and DSR is very high in deliver packets when the number of packets and connec-
Random Walk and Random Direction, but its not so high tions are increased. AODV spends much lesser time than
in Random waypoint. other protocols under random walk and Random direction
mobility models
1500
800
AODV 10
1200
Avg. Delay in m/s

AODV 20
AODV 10
DSDV 10 600 AODV 20
Avg. Delay in m/s

900
DSDV 20 DSDV 10
TORA 10 DSDV 20
600 400
TORA 20 TORA 10
DSR 10 TORA 20
300 DSR 20 200 DSR 10
DSR 20
0
10 20 0
Speed in m/s 1 4
Packets per second

a) Random way point

a) Random way point
2400
4000
2100 AODV 10 3500
Avg. Delay in m/s

1800 AODV 20 AODV 10

3000 AODV 20
Avg. Delay in m/s

1500 DSDV 10
2500 DSDV 10
DSDV 20
1200 DSDV 20
TORA 10 2000
TORA 10
900 TORA 20
1500 TORA 20
600 DSR 10
1000 DSR 10
DSR 20
300 DSR 20
500
0
0
10 20 1 4
Speed in m/s Packets per second

b) Random Walk
b) Random Walk
28 IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009

2000 1400
1800
AODV 10 1200
1600 AODV 10
AODV 20
Avg. Delay in m/s

AODV 20

Avg. Delay in m/s

1400 1000
DSDV 10 DSDV 10
1200
DSDV 20 800 DSDV 20
1000
TORA 10 TORA 10
800 600
TORA 20 TORA 20
600
DSR 10 400 DSR 10
400 DSR 20
DSR 20
200 200
0
0
1 4
400x400 1000x600
Packets per seconds
Simulation Area

c) Random Direction c) Random Direction

Figure 8. End-to-end delay for Traffic load Figure 9. End-to-end delay for node Density

4.6 Node Density vs End-to-End Delay 5. Discussion

The performance of the routing protocols in terms of end- In Random Waypoint model, most of the times the nodes
to-end delay is examined with respect to the area with in choose destination closer to the centre of the simulation
which the nodes are likely to move.. Two traffic density area and thus producing a dense wave near the centre and
scenarios are considered- one with 10 connections and stays back there for the specified pause time, also having
another with 20 connections. The results are shown graph- more neighbors to the nodes in the centre. This will give
ically in figure 9. minimal hop distance between the source-destination pairs.
The end-to-end delay is very less with higher node density When the network becomes sparse or the traffic load be-
and increases heavily when the node becomes sparse. For comes high the performance produced by DSR and TORA
the varying node density the end-to-end delay produced by decreases sharply.. DSDV protocol’s performance is
the protocols in Random waypoint is very less and very nearer to AODV under network size metric. TORA
high in Random walk and Random Direction Model. protocol’s performance was not so good under this
AODV in Random Way point model Performs better than mobility model.
other mobility models. The Random Walk model creates a high mobility sce-
nario with larger travel time the nodes will travel almost to
1500
all the areas. Since there is no pause time between change
1200 AODV 10 of speed and direction, the need for a protocol that updates
AODV 20
Avg. Delay in m/s

DSDV 10
the routing information quickly as uses the fresh informa-
900
DSDV 20 tion about the routing becomes mandatory. The simulation
600
TORA 10 results show that the AODV performs better than DSR,
TORA 20
DSR 10 TORA and DSDV. One of the reason here is the average
300 DSR 20 hop distance between the source-destination becomes high,
0
and this will increase packet overhead. The usage of the
400x400 1000x600 fresh route information and quickly adapting nature of
Simulation Area
AODV are reasons for better results produced by the
AODV. DSDV produces better results than TORA and
a) Random way point can be used as the routing protocol under low mobili-
ty conditions.
2700
2400
The Random Direction Model is an unrealistic model
AODV 10
2100 AODV 20
because it is unlikely that people would spread them-
Avg. Delay in m/s

1800 DSDV 10 selves evenly throughout an area. The nodes choose

1500 DSDV 20 pause times only at the boundaries and no change of speed
1200 TORA 10
TORA 20
and direction before reaching the boundary. This will
900
600 DSR 10 create a topography in which most of the times most of the
300
DSR 20 nodes are in the boundary and the centre of the area be-
0 comes very sparse. Here the average number of hop dis-
400x400 1000x600
Sim ulation Area tance becomes higher and gives lesser number of alterna-
tive paths. AODV protocol produces better results than
DSDV, TORA and DSR. When the network size is
b) Random Walk large, DSDV produces better results than TORA and
DSR.
IJCSNS International Journal of Computer Science and Network Security, VOL.9 No.2, February 2009 29

6. Conclusion on Mobile Computing Systems and Applica-

tions(WMCSA ’99), New Orleans, pp 90-100,
In Random way point model the simulation results shows February 1999.
that when the network becomes sparse or the traffic load [6] Perkins C.E. and P. Bhagwat, “Highly Dynamic
becomes high the performance produced by DSR and TO- Destination-Sequenced Distance Vector Routing
RA decreases sharply. DSDV protocol’s performance is (DSDV) for Mobile Computers”, Computer Com-
closer to AODV under network size metric. TORA munications Review, pages 234-244, October 1994.
protocol’s performance was not so good under this [7] Samir. R. Das, R. Castaneda and J.Yan. “Simula-
mobility model. Hence, AODV protocol can be cho- tion based Performance Evaluation of Routing
sen as the routing protocol in this type of mobility Protocols for Mobile Ad-hoc Net-
conditions. works”ACM/Baltzer Mobile Networks and Appli-
In random walk model, AODV performs better than cations (MONET), pp 179-189, 2000.
DSR, TORA and DSDV because the average hop dis- [8] Samir R. Das, Robert Castaneda, Jiangtao Yan,
tance between the source-destination becomes high in Rimli Sengupta, “Comparative Performance
AODV, and this will increase packet overhead. So AODV Evaluation of Routing Protocols for Mobile Ad
protocols perform better under low and high mobility Hoc Networks”, 7th International Conference on
conditions. on Computer Communication and networks( IC3N,
The Random Direction Model produces better results pp 153-161, October 1998.
than DSDV, TORA and DSR. When the network size [9] Tony Larsson, Nicklar Hedman, “Routing Proto-
is large, DSDV produces better results than TORA cols in Wireless Ad Hoc Networks – A Simulation
and DSR. This shows that AODV is the suitable choice Study”, Masters thesis in Computer Science and
under this mobility model. Engineering, Stockholm, Lulea University of
In this paper, only four ad-hoc routing protocols were Technology, 1998.
considered and their performance were analyzed only [10] Tracy Camp, Jeff Boleng, Vanessa Davies, “A
under the Random based mobility models. In future, Survey of Mobility Models in Wireless Ad-hoc
this paper can be enhanced by analyzing the other ad- Networks” Wireless Communications & Mobile
hoc routing protocols under real-world scenarios such Computing (WCMC): Special issue on Mobile Ad
as Group-mobility models. Hoc Networking: Research, Trends and Applica-
tions, vol. 2, no. 5, pp. 483-502, April 2002.
[11] Vanessa Ann Davies, “A Thesis on “Evaluating
7. REFERENCES Mobility Models within an ad-hoc networks”, Co-
laroda School of Mines, 2000.
[1] Amir R.Das, Charles E.Perkins and Elizabeth [12] Vincent D Park and M Scott Corson, “A Highly
M.Royer. “An Implementation Study of the Adaptive Distributed Routing Algorithm for Mo-
AODV routing protocol”, Proceedings of the bile Wireless Networks”, Proceedings of IEEE
IEEE Wireless Communication and Networking Conference INFOCOM’97 on Computer Communi-
Conference(INFOCOM), Tel Aviv, Isrel, pp 3-12, cations, pp. 1405-1413, April 7-12, 1997.
March 2000.
[2] Josh Broch, David A.Maltz, David B. Johnson
Yih-Chen Hu and Jorjeta Jetcheva, “A Perfor-
mance Comparison of Multihop Wireless
Ad Hoc Network Routing Protocols”, ACM MO-
BICOM 98, Dallas, Texas. pp 25-30, Octo-
ber 1998.
[3] Jochen Schiller “Mobile Communications”, Ad-
dision Wesley Longman Pvt.Ltd, India. 2000.
[4] Johanson P, Larsson.T, Hedman N, Mielczarek
and degrermark M., “Routing Protocols for Mobile
Ad Hoc Networks – a comparative performance
analysis”, Proceedings of the ACM/IEEE Interna-
tional Conference on Mobile Computing and Net-
working(MOBICOM), Seattle, WA, pp 195-206,
august 1999.
[5] Perkins C. and Royer.E. “Ad Hoc On-Demand
Distance Vector Routing”. The 2nd IEEE workshop