Hindawi Publishing Corporation

Journal of Computer Systems, Networks, and Communications

Volume 2008, Article ID 860364, 10 pages
doi:10.1155/2008/860364

Research Article
Reference Point Group Mobility and Random Waypoint Models
in Performance Evaluation of MANET Routing Protocols

Geetha Jayakumar and Gopinath Ganapathi

Department of Computer Sciences, Bharathidasan University, Tiruchirappalli 620 023, Tamil Nadu, India

Correspondence should be addressed to Geetha Jayakumar, [email protected]

Received 7 August 2008; Revised 29 September 2008; Accepted 5 December 2008

Recommended by Walaa Hamouda

Ad hoc networks are characterized by multihop wireless connectivity, frequently changing network topology and the need for
efficient dynamic routing protocols plays an important role. We compare the performance of two prominent on-demand routing
protocols for mobile ad hoc networks: dynamic source routing (DSR), ad hoc on-demand distance vector routing (AODV). A
detailed simulation model with medium access control (MAC) and physical layer models is used to study the interlayer interactions
and their performance implications. We demonstrate that even though DSR and AODV share similar on-demand behavior, the
differences in the protocol mechanisms can lead to significant performance differentials. In this paper, we examine both on-
demand routing protocols AODV and DSR based on packet delivery ratio, normalized routing load, normalized MAC load, average
end-to-end delay by varying the node density, network loading, and mobility variations for reference point group mobility and
random waypoint models. This framework aims to evaluate the effect of mobility models on the performance of mobile ad hoc
networks (MANETs) routing protocols. Our results show that the protocol performance may vary drastically across mobility
models and performance rankings of protocols may vary with the mobility models used. This effect can be explained by the
interaction of the mobility characteristics with the connectivity graph properties.

Copyright © 2008 G. Jayakumar and G. Ganapathi. This is an open access article distributed under the Creative Commons
Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.

1. INTRODUCTION wired and wireless networks to suit today’s need of mobile

communication. As the mobility of users continues to
Wireless technologies are without any doubt shaping the increase, a special type of network will be gaining more
beginning of the new millennium. Cellular networks (2/2.5/3 and more attention which is mobile ad hoc network. Such
Generations) have reached global coverage and wireless research on the capacity of wireless ad hoc networks and
LANs are providing high speed internet access to city center the popular IEEE 802.11 protocol typically assume the use
and hot spot areas. The principle of computing anytime of omnidirectional antennas at all nodes [2]. Over the
anywhere is becoming reality. Consequently, there has been past few years, considerable amount of research addressed
tremendous interest recently in the convergence of wired the issues, such as, (a) access, (b) routing, (c) transport
(fixed) networks and mobile wireless and local wireless layer protocols, (d) quality of service provisioning, (e) self-
access systems and networks. Beyond 3G (B3G) and 4G organization, and (f) energy management. Recent research
concepts have been the very hot subject of much research includes understanding the cross-layer protocol interactions
and many standardization activities throughout the world. and new designs of such protocols [3, 4].
Many B3G/4G-related technical committees are taking the In a mobile ad hoc network, nodes do not rely on
initiative to produce enhanced proposals of radio trans- any existing infrastructure. Instead, the nodes themselves
mission technologies and wireless applications protocols form the network and communicate through means of
[1]. wireless communications. Mobility causes frequent topology
Computer network, traditionally viewed as infrastruc- changes and may break existing paths. A routing protocol
ture of a fixed form, has evolved into combinations of should quickly adapt to the topology changes and efficiently
2 Journal of Computer Systems, Networks, and Communications

search for new paths. On the other hand, the limited power It is necessary to systematically propose a framework to
and bandwidth resources in mobile ad hoc networks make analyze the impact of mobility on the performance of routing
quick adaptation very challenging [5]. Nodes will have to protocols in ad hoc networks. Through this framework, we
forward the network traffic on behalf of other nodes to illustrate how modeling mobility is important in affecting
allow communication to take place between nodes that are routing performance and understanding the mechanism of
out of each other immediate radio range. Hence, routing of ad hoc routing protocols. Our framework mainly focuses
network traffic becomes a central issue in these networks. on the following aspects: mobility models, the metrics for
The mobile ad hoc networks are envisioned to support mobility and connectivity graph characteristics, the potential
dynamic and rapidly changing multihop topologies which relationship between mobility and routing performance, and
are likely to be composed of relatively bandwidth constrained the analysis of impact of mobility on building blocks of ad
wireless links [6]. A generic framework to systematically hoc routing protocols.
analyze the impact of mobility on the performance of routing
protocols for MANET has become important. It is necessary 2.1. Reactive protocols
to find out what degree of mobility affects routing protocol
performance. Many previous studies have used random In contrast to table-driven routing protocols, all up-to-
waypoint (RWP) as reference model [7]. Random waypoint date routes are not maintained at every node; instead the
is a simple model that is easy to analyze and implement. This routes are created as and when required. When a node
has probably been the main reason for the widespread use requires a route to the destination, it invokes a route
of this model for simulations. Further, the RWP model is discovery mechanism to find the path to the destination. The
not sufficient to capture some realistic scenarios. In order to process is completed once a route is found or all possible
model the movements of nodes in a realistic terrain, such as route permutations have been examined. Once a route has
a battlefield or rescue operation, some sophisticated mobility been established, it is maintained by a route maintenance
models such as reference point group mobility (RPGM) procedure until either the destination becomes inaccessible
models find valuable application. along every path from the source or until the route is
In the RWP model, the nodes, that is, mobile users, no longer desired. This section discusses two on-demand
move along a zigzag path consisting of straight legs from one routing protocols.
waypoint to the next [8]. Mobility model such as RWP model
is described on graphs based on road maps [9]. 2.2. AODV
For this purpose, numerous ad hoc routing protocols
have been proposed by the Internet Engineering Task Force AODV (see [11]) is an improvement on the destination-
(IETF) working group MANET. These protocols have been sequenced distance vector (DSDV). AODV uses an on-
evaluated through simulations in network simulators like demand approach for finding routes. Since it is an on-
NS-2. demand algorithm, a route is established only when it is
required by a source node for transmitting data packets and
2. BACKGROUND it maintains these routes as long as they are needed by the
sources. AODV uses a destination sequence number, created
The routing protocols for MANET can be broadly classified by the destination, to determine an up-to-date path to the
as on-demand/reactive and periodic/proactive protocols. destination. A node updates its route information only if the
Reactive routing protocols propagate route updates only destination sequence number of the current received packet
when a route to a destination is required. There are several is greater than the destination sequence number stored at the
reactive routing protocols available for ad hoc networks, node. It indicates the freshness of the route accepted by the
including dynamic source routing (DSR) [10], ad hoc on- source. To prevent multiple broadcast of the same packet,
demand distance vector (AODV) [11], and so on. Reactive AODV uses broadcast identifier number that ensures loop
routing protocols have been demonstrated to perform better freedom since the intermediate nodes only forward the first
with significantly lower overheads than proactive routing copy of the same packet and discard the duplicate copies.
protocols in many scenarios [12] since they are able to To find a path to the destination, the source broadcasts a
react quickly to topology changes, yet being able to reduce route request (RREQ) packet across the network. This RREQ
routing overhead in periods or areas of the network in which contains the source identifier, the destination identifier, the
changes are less frequent. In this section, we briefly discuss source sequence number, the destination sequence number,
the working of both on-demand routing protocols AODV the broadcast identifier, and the time to live field. Nodes that
and DSR. Their respective performances are compared in receive the RREQ find out whether they are the destination
Section 1. It has also been observed that under a given or whether they have a fresh route to the destination, then
mobility pattern, routing protocols like DSR and AODV they respond to the RREQ by unicasting a route reply (RREP)
perform differently. back to the source node.
This is possibly because each protocol differs in the basic Otherwise, the node rebroadcasts the RREQ. When a
mechanisms or “building blocks” it uses. For example, DSR node forwards an RREQ packet to its neighbors, it also
uses route discovery. To find out how, we need to investigate records in its tables the node from which the first copy
the effect of mobility on some of these “building blocks” of the request came. This information is used to construct
and how they impact the protocol performance as a whole. the reverse path for the RREP packet. AODV uses only
G. Jayakumar and G. Ganapathi 3

symmetric links because the route reply packet follows node can use an alternate route from its own cache, when
the reverse path of route request packet. When a node a data packet meets a failed link on its source route. (ii)
receives an RREP packet, information about the previous Gratuitous route repair: a source node receiving an RERR
node from which the packet was received is also stored packet piggybacks the RERR in the following RREQ. This
in order to forward the data packets to this next node as helps cleaning up the caches of other nodes in the network
the next hop toward the destination. Once the source node that may have the failed link in one of the cached source
receives an RREP, it can begin using the route to send data routes. (iii) Promiscuous listening: when a node overhears a
packets. The source node rebroadcasts the RREQ if it does packet not addressed to itself, it checks if the packet could be
not receive an RREP before the timer expires. It attempts routed via itself to gain a shorter route. If so, the node sends
discovery up to some maximum number of attempts. If it a gratuitous RREP to the source of the route with this new,
does not discover a route after this maximum number of better route. Aside from this, promiscuous listening helps a
attempts, the session is aborted. If the source moves then it node to learn different routes without directly participating
can reinitiate route discovery to the destination. If one of the in the routing process.
intermediate nodes moves then the moved node neighbor
realizes the link failure and sends a link failure notification 3. CLASSIFICATION OF MOBILITY MODELS
to its upstream neighbors and so on till it reaches the
source upon which the source can reinitiate route discovery In general, the mobility models can be classified according to
if needed. The main advantage of AODV is that routes are the different kinds of dependencies and restrictions that are
obtained on demand and destination sequence numbers are considered.
used to find the latest route to the destination. One of
the disadvantages of AODV is that intermediate nodes can (i) Random based: there are neither dependencies nor
lead to inconsistent routes if the source sequence number any other restrictions modeled which are similar to
is very old and the intermediate nodes have a higher but RWP model.
not the latest destination sequence number, thereby causing (ii) Temporal dependencies: the actual movement of a
stale entries. Also multiple route reply (RREP) packets in node is influenced by the movement of the past.
response to a single route request (RREQ) packet can lead
(iii) Spatial dependencies: the movement of a node is
to heavy control overhead. Periodic hello messages also lead
influenced by the nodes around it, for example, group
to unnecessary bandwidth consumption.
mobility such as RPGM model.

2.3. DSR (iv) Geographic restrictions: the area in which the node is
allowed to move is restricted.
The key feature of DSR [10, 13] is the use of source routing. (v) Hybrid characteristics: a combination of temporal
That is, the sender knows the complete hop-by-hop route dependencies, spatial dependencies, and geographic
to the destination. These routes are stored in a route cache. restrictions is realized [15].
The data packets carry the source route in the packet header.
When a node in the ad hoc network attempts to send a data
4. MODELS
packet to a destination for which it does not already know
the route, it uses a route discovery process to dynamically Random waypoint model
determine such a route. Route discovery works by flooding
the network with route request (RREQ) packets. Each node The random waypoint mobility model is simple and is widely
receiving an RREQ rebroadcasts it, unless it is the destination used to evaluate the performance of MANETs. The random
or it has a route to the destination in its route cache. Such a waypoint mobility model contains pause time between
node replies to the RREQ with a route reply (RREP) packet changes in direction and/or speed. Once a mobile node
that is routed back to the original source. RREQ and RREP (MN) begins to move, it stays in one location for a specified
packets are also source routed. The RREQ builds up the path pause time. After the specified pause time is elapsed, the
traversed so far. The RREP routes itself back to the source by MN randomly selects the next destination in the simulation
traversing this path backwards, the route carried back by the area and chooses a speed uniformly distributed between the
RREP packet is cached at the source for future use. If any link minimum speed and maximum speed and travels with a
on a source route is broken, the source node is notified using speed v whose value is uniformly chosen in the interval
a route error (RERR) packet. The source removes any route (0, Vmax ). Vmax is some parameter that can be set to reflect
using this link from its cache. A new route discovery process the degree of mobility. Then, the MN continues its journey
must be initiated by the source if this route is still needed. toward the newly selected destination at the chosen speed. As
DSR makes very aggressive use of source routing and route soon as the MN arrives at the destination, it stays again for
caching. No special mechanism to detect routing loops is the indicated pause time before repeating the process [16].
needed. Also, any forwarding node caches the source route in
a packet it forwards for possible future use. Several additional Reference point group mobility model (RPGM)
optimizations have been proposed and have been evaluated
to be very effective by the authors of the protocol [14], as This model is described as another way to simulate group
described in the following. (i) Salvaging: an intermediate behavior in [17], where each node belongs to a group
4 Journal of Computer Systems, Networks, and Communications

where every node follows a logical center (group leader) (i) In-place mobility model: the entire field is divided into
that determines the group’s motion behavior. The nodes in a several adjacent regions. Each region is exclusively
group are usually randomly distributed around the reference occupied by a single group. One such example is
point. The different nodes use their own mobility model and battlefield communication.
are then added to the reference point which drives them in (ii) Overlap mobility model: different groups with differ-
the direction of the group. At each instant, every node has ent tasks travel on the same field in an overlapping
a speed and direction that is derived by randomly deviating manner. Disaster relief is a good example.
from that of the group leader. This general description of
group mobility can be used to create a variety of models (iii) Convention mobility model: this scenario is to emulate
for different kinds of mobility applications. Group mobility the mobility behavior in the conference. The area is
as such can be used in military battlefield communications. also divided into several regions while some groups
One example of such mobility is that a number of soldiers are allowed to travel between regions.
may move together in a group. Another example is during
disaster relief where various rescue crews (e.g., firemen, In RPGM model, the vector RMi indirectly determine how
policemen, and medical assistants) form different groups and much the motion of group members deviate from their
work cooperatively. The respective functions of group leaders leader. The movement can be characterized as follows:
   
and group members are described as follows. Vmember (t) = Vleader (t) + random()∗SDR∗max speed,

θmember (t) = θleader (t) + random()∗SDR∗max angle

4.1. The group leader (2)
The movement of group leader at time, t, can be represented
t (see [7]), where 0 < SDR, ADR < 1. SDR is the speed
by motion vector Vgroup . Not only it defines the motion of
deviation ratio and ADR is the angle deviation ratio. SDR
group leader itself, but also it provides the general motion
and ADR are used to control the deviation of the velocity
trend of the whole group. Each member of this group
t (magnitude and direction) of group members from that
deviates from this general motion vector Vgroup by some
t
of the leader. By simply adjusting these two parameters,
degree. The motion vector Vgroup can be randomly chosen different mobility scenarios can be generated.
or carefully designed based on certain predefined paths [18]. Because of the inherent characteristic of spatial depen-
dency between nodes, the RPGM model is expected to
4.2. The group members behave differently from the RWP model. We find that RPGM
incurs less link breakage and achieves a better performance
The movement of group members is significantly affected for various routing protocols than RWP model.
by the movement of its group leader. For each node, We observe that the mobility models have various
mobility is assigned with a reference point that follows the properties and exhibit different mobility characteristics. As
group movement. Upon this predefined reference point, each a consequence, we expect that mobility models behave dif-
mobile node could be randomly placed in the neighborhood. ferently and influence the protocol performance in different
Formally, the motion vector of group member, i, at time ways. Therefore, to thoroughly evaluate ad hoc protocol
t, Vit , can be described as performance, it is imperative to use a rich set of mobility
models. Each model has its own unique and specific mobility
characteristics. Hence, a method to choose a suitable set of
Vit = Vgroup
t
+ RMti , (1) mobility models is needed.

4.3. Simulation and performance metrics

where the motion vector RMti is a random vector deviated
by group member i from its own reference point. The For our scenario-based experiments, we used the ns-2 simu-
vector RMti is an independent identically distributed ran- lator which is available as an open source distribution [19].
dom process whose length is uniformly distributed in the For generating the scenarios, we used the mobility scenario
interval [0, rmax ] (where rmax is maximum allowed distance generation tool, BonnMotion. The radio model corresponds
deviation) and whose direction is uniformly distributed in to the 802.11 WaveLAN, operating at a maximum airlink rate
the interval [0, 2π]. of 2 Mbps. The media access control protocol used is the
t
Vgroup is the motion vector for the group leader; it is also IEEE 802.11 distributed coordination function (DCF). The
the motion vector for the whole group. RMti is the random traffic pattern file was generated using “cbrgen.tcl” script,
deviation vector for group member i, and the final motion which is provided along with the standard ns-2 distribution.
vector of group member i is represented by vector Vit . We used constant bit rate (CBR) traffic for our simulations.
With appropriate selection of predefined paths for group
leader and other parameters, the RPGM model is able to 4.4. Metrics
emulate a variety of mobility behaviors. RPGM model is
able to represent various mobility scenarios including the We have selected the packet delivery ratio, average end-to-
following. end delay, and protocol control overhead as a metric during
G. Jayakumar and G. Ganapathi 5

Table 1: Parameters for RPGM and random waypoint model-mobility variations.

Parameters Values Values

Mobility model RPGM Random waypoint
Simulation area 1000 m × 1000 m 1000 × 1000
Minimum node speed 1 m/s Uniform speed
# of nodes 20 20
Traffic type CBR (constant bit rate) CBR
Packet size 512 bytes 512 bytes
Connection rate 10 pkts/s 10 pkts/s
Pause time 25 s 25 s
Number of connections 5 5
Maximum node speed 10–15–20–25 m/s 10–15–20–25 m/s
Simulation time 200 s 200
Distribution of nodes 2 in each group 10 groups —
Protocols AODV, DSR AODV, DSR
Probability of group change 0.05 —
Maximum distance to group center 100 m —
Standard deviation 2.0 —

Table 2: Parameters for RPGM and random waypoint-node density variations.

Parameters Values Values

Mobility model RPGM Random waypoint
Simulation area 1000 m × 1000 m 1000 m × 1000 m
Minimum node speed 1 m/s Uniform speed
Number of nodes 20, 50, 80, 100 20, 50, 80, 100
Traffic type CBR (constant bit rate) CBR
Packet size 512 bytes 512 bytes
Connection rate 20 pkts/s 20 pkts/s
Pause time 20–30 s 25 s
Number of connections 5 5
Maximum node speed 10 m/s 10 m/s
Simulation time 200 s 200 s
Protocols AODV, DSR AODV, DSR
Distribution of nodes 2 in each group 10 groups —
Probability of group change 0.05 —
Maximum distance to group center 100 m —
Standard deviation 2.0 —

the simulation in order to evaluate the performance of the (b) Normalized routing load (NRL): this is calculated
different protocols [20]. as the ratio between the number of routing pack-
ets transmitted to the number of packets actually
(a) Packet delivery fraction (PDF): this is the ratio of
received (thus accounting for any dropped packets):
total number of packets successfully received by the
destination nodes to the number of packets sent by number of routing packets sent
the source nodes throughout the simulation: NRL = . (4)
number of data packets received
number of received packets
PDF = . (3) This metric gives an estimate of how efficient a
number of sent packets
routing protocol is since the number of routing
This estimate gives us an idea about how successful packets sent per data packet gives an idea of how
the protocol is in delivering packets to the application well the protocol maintains the routing information
layer. A high value of PDF indicates that most of the updated. The higher the NRL, the higher the over-
packets are being delivered to the higher layers and is head of routing packets and consequently the lower
a good indicator of the protocol performance. the efficiency of the protocol is.
6 Journal of Computer Systems, Networks, and Communications

PDF versus mobility (m/s) NRL versus mobility (m/s)

120 1.2

100 1

80 0.8
PDF (%)

NRL
60 0.6

40 0.4

20 0.2

0 0
0 10 20 30 0 10 20 30
Mobility (m/s) Mobility (m/s)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(a) (b)

NMAC versus mobility (m/s) End to end delay versus mobility (m/s)
18 1.6
16 1.4
14 End to end delay (s)
1.2
12
1
NMAC

10
0.8
8
0.6
6
4 0.4
2 0.2
0 0
0 10 20 30 0 10 20 30
Mobility (m/s) Mobility (m/s)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(c) (d)

Figure 1

Table 3: Parameters for RPGM and random waypoint-network loading.

Parameters Values Values

Mobility model RPGM Random waypoint
Simulation area 1000 m × 1000 m 1000 m × 1000 m
Minimum node speed 1 m/s Uniform speed
Number of nodes 20 20
Traffic type CBR (constant bit rate) CBR
Packet size 512 bytes 512 bytes
Connection rate 20–30–40–50–60 pkts/s 20–30–40–50–60 pkts/s
Pause time 20–30 s 25 s
Number of connections 5 5
Maximum node speed 10 m/s 10 m/s
Simulation time 200 s 200 s
Distribution of nodes 2 in each group 10 groups —
Probability of group change 0.05 —
Maximum distance to group center 100 m —
Standard deviation 2.0 —
G. Jayakumar and G. Ganapathi 7

PDF versus node density NRL versus node density

120 3.5

100 3

2.5
80
PDF (%)

NRL
60
1.5
40
1
20 0.5

0 0
0 50 100 150 0 50 100 150
Node density (number of nodes per area) Node density (number of nodes per area)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(a) (b)

NMAC versus node density End to end delay versus node density
50 4
3.5
40 End to end delay (s)
3

30 2.5
NMAC

2
20 1.5
1
10
0.5
0 0
0 50 100 150 0 50 100 150
Node density (number of nodes per area) Node density (number of nodes per area)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(c) (d)

Figure 2

(c) Average end-to-end delay (AED): this is defined as the The scenarios depict varying node densities, link changes,
average delay in transmission of a packet between two and mobility variations. To thoroughly study the effect of
nodes and is calculated as follows: mobility on MANET protocol performance, we need to
evaluate the protocols over a rich set of mobility models that
n 


span the design space of the mobility metrics. The scenarios
time packet receivedi − time packet senti
AED = . are explained in the following sections.
i=0
total number of packets received
The reference point group mobility (RPGM) model [21] is
(5)
used for modeling the scenario. In RPGM model, a cluster of
nodes communicates in groups. The velocity and direction of
A higher value of end-to-end delay means that the nodes within the group are determined by a “group leader”
network is congested and hence the routing protocol or reference point. Since the group leader mainly decides
does not perform well. The upper bound on the the mobility of group members, group mobility pattern is
values of end-to-end delay is determined by the expected to have high spatial dependence for small values
application. of SDR and ADR. We define the parameters in this mobility
model as shown in Table 1.
5. DESCRIPTION OF THE SCENARIOS In random waypoint model, a mobile node chooses a
random destination at every instance and moves toward it
We consider different scenarios for our experiments, in with a speed uniformly distributed [0, Vmax ], where Vmax is
which the nodes are distributed over the simulation area. the maximum allowable speed for a node. After reaching the
8 Journal of Computer Systems, Networks, and Communications

PDF versus network loading NRL versus network loading

100 0.7

0.6
80
0.5
60
PDF (%)

0.4

NRL
40 0.3

0.2
20
0.1

0 0
0 20 40 60 80 0 20 40 60 80
Network loading (pkts/s) Network loading (pkts/s)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(a) (b)

NMAC versus network loading End to end delay versus network loading
14 1.6
12 End to end delay (s) 1.4
1.2
10
1
NMAC

8
0.8
6
0.6
4
0.4
2 0.2
0 0
0 20 40 60 80 0 20 40 60 80
Network loading (pkts/s) Network loading (pkts/s)

RPAODV RWAODV RPAODV RWAODV

RPDSR RWDSR RPDSR RWDSR
(c) (d)

Figure 3

destination, the node stops for a duration defined by the like route requests (RREQs) and hence are less expensive.
“pause time” parameter. After this duration, it again chooses Consequently, when the MAC overhead is factored, DSR is
a random destination and repeats the whole process again found to generate higher overall network load than AODV
until the simulation ends. in all scenarios despite having less routing overhead. The
end-to-end delay of AODV is less when compared to DSR
(Figure 1(d)).
5.1. Mobility variations results

The analysis of Figure 1(a) shows that AODV and DSR 5.2. Node density results
have similar performance outputs. From Figure 1(b), we
infer that routing overheads generated by AODV is greater In the second experiment on RPGM model, we try to find
than DSR. This can be attributed to more routing updates the performance aspect when the node density is varied
needed in AODV. DSR uses source routing and also caches within a fixed map area. The node density is incrementally
some routing entries. The normalized medium access control adjusted from 20–100 nodes within the same map area of
(MAC) load for AODV is found to be less when compared to 1000 × 1000 m. The simulation parameters are shown in
DSR. This is because route errors (RERRs) (Figure 1(c)) are Table 2. From Figure 2(a), we infer that DSR and AODV have
handled differently in each protocol. RERRs are unicast in similar results. The routing overhead remains relatively low
DSR and therefore contribute to additional MAC overhead for DSR whereas for AODV it increases tremendously. The
like route replies (RREPs). In AODV, RERRs are broadcast network delay of AODV is less when compared to DSR.
G. Jayakumar and G. Ganapathi 9

5.3. Network loading results used. This clearly shows that mobility pattern influ-
ences the connectivity graph which in turn influences
In the third set of experiments of RPGM, we try to find the protocol performance.
the routing performance aspect when the offered load
increases. To do this, we increase the average connection
load offered by each connection starting at 20–60 pkts/s. ACKNOWLEDGMENT
The same parameters are used, such as the type of traffic The authors would like to thank the anonymous reviewers
is constant bit rate (CBR) and the nodes are pumping for their careful reading and insightful comments that have
data as shown in Table 3. Figure 3(a) shows that the PDF helped in improving the presentation of this paper.
of routing protocols have very close performance results
when the network loading is high. Overall, the protocols
experience a drop in packet delivery ratio as network loading REFERENCES
increases. Figure 3(b) shows that DSR has less overhead
[1] I. Awan and K. Al-Begain, “Performance evaluation of wireless
when compared to AODV. Figure 3(c) shows that DSR has networks,” International Journal of Wireless Information Net-
a high-normalized MAC load when compared to AODV. works, vol. 13, no. 2, pp. 95–97, 2006.
We see from Figure 3(d) that AODV has lowest network [2] S. Yi, Y. Pei, S. Kalyanaraman, and B. Azimi-Sadjadi, “How
latency in the RPGM model when network loading increases is the capacity of ad hoc networks improved with directional
compared to DSR. antennas?” Wireless Networks, vol. 13, no. 5, pp. 635–648,
2007.
6. CONCLUSION [3] S. L. Kota, E. Hossain, R. Fantacci, and A. Karmouch, “Cross-
layer protocol engineering for wireless mobile networks—part
(1) In this study, analysis has been made on the impact of 1,” IEEE Communications Magazine, vol. 43, no. 12, pp. 110–
mobility pattern on routing performance of mobile 111, 2005.
ad hoc network in a systematic manner. [4] S. L. Kota, E. Hossain, R. Fantacci, and A. Karmouch, “Cross-
layer protocol engineering for wireless mobile networks—part
(2) It has been observed that the mobility pattern influ- 2,” IEEE Communications Magazine, vol. 44, no. 1, pp. 83–84,
ences the performance of MANET routing protocols. 2006.
There is a very clear trend between mobility metric, [5] D. R. Grinath and S. Selvan, “Performance analysis of disper-
connectivity, and performance. sion mobility model in mobile ad hoc networks,” International
Journal of Computer Science and Network Security, vol. 8, no. 3,
(3) It has been observed that with similar average pp. 77–87, 2008.
spatial dependency when the relative speed increases, [6] R. Folio, J. B. Cain, and S. Kota, “Challenges in the verification
the link duration decreases and hence the routing of mobile ad hoc networking systems,” International Journal
overhead increases and throughput decreases. of Wireless Information Networks, vol. 14, no. 2, pp. 107–120,
(4) With similar average relative speed, it is found that 2007.
[7] G. Lin, G. Noubir, and R. Rajaraman, “Mobility models for
spatial dependence increases and the link duration
ad hoc network simulation,” in Proceedings of the 23rd Annual
increases and therefore, the throughput increases and
Joint Conference of the IEEE Computer and Communications
the routing overhead decreases. Societies (INFOCOM ’04), vol. 1, pp. 454–463, Hongkong,
(5) It has been observed that DSR and AODV achieve March 2004.
the highest throughput and least overhead with [8] E. Hyytiä and J. Virtamo, “Random waypoint mobility model
RPGM when compared to RWP mobility models. in cellular networks,” Wireless Networks, vol. 13, no. 2, pp. 177–
This is because with similar relative speed, between 188, 2007.
random waypoint and RPGM, high degree of spatial [9] P. S. Mogre, M. Hollick, N. d’Heureuse, H. W. Heckel, T.
Krop, and R. Steinmetz, “A graph-based simple mobility
dependence for RPGM means higher link duration
model,” in Proceedings of the 4th Workshop on Mobile Ad-
and correspondingly higher path duration, which Hoc Networks (WMAN ’07), pp. 421–432, Bern, Switzerland,
in turn will result in higher throughput and lower February-March 2007.
routing overhead. [10] D. B. Johnson and D. Maltz, “Dynamic source routing in ad
(6) RWP model is insufficient to capture the following hoc wireless networks,” in Mobile Computing, T. Imielinski
mobility characteristics such as temporal dependence and H. Korth, Eds., pp. 153–181, Kluwer Academic Publishers,
Dordrecht, The Netherlands, 1996.
and spatial dependence. This reduces the throughput
[11] C. E. Perkins and E. M. Royer, “Ad hoc on-demand distance
when compared to RPGM.
vector routing,” in Proceedings of the 2nd IEEE Workshop on
(7) From the results, it is analyzed that AODV has better Mobile Computing Systems and Applications (WMCSA ’99), pp.
throughput and less delay in RPGM model when 90–100, New Orleans, La, USA, February 1999.
compared to RWP model. For circumstances such as [12] J. Broch, A. D. Maltz, D. B. Johnson, Y.-C. Hu, and J. G.
military operation, AODV may emerge as a better Jetcheva, “A performance comparison of multi-hop wireless
choice. ad hoc network routing protocols,” in Proceedings of the
4th Annual ACM/IEEE International Conference on Mobile
(8) Thus, it is necessary to conclude that relative rankings Computing and Networking (MobiCom ’98), pp. 85–97, Dallas,
of the protocols may vary with the mobility model Tex, USA, October 1998.
10 Journal of Computer Systems, Networks, and Communications

[13] J. Broch, D. Johnson, and D. Maltz, “The dynamic source

routing protocol for mobile ad hoc networks,” IETF Internet
Draft, December 1998, http://www.ietf.org/internet-drafts/
draft-ietf-manet-dsr-11.txt.
[14] D. A. Maltz, J. Broch, J. Jetcheva, and D. B. Johnson,
“The effects of on-demand behavior in routing protocols for
multihop wireless ad hoc networks,” IEEE Journal on Selected
Areas in Communications, vol. 17, no. 8, pp. 1439–1453, 1999.
[15] N. Aschenbruck, E. Gerhards-Padilla, and P. Martini, “A
survey on mobility models for performance analysis in
tactical mobile networks,” Journal of Telecommunications and
Information Technology, vol. 2, pp. 54–61, 2008.
[16] I. Khider, W. Furong, and Y. Wei, “Study on indoor and
outdoor environment for mobile ad hoc network supported
with base sstations,” in Proceedings of the International Con-
ference on Wireless Communications, Networking and Mobile
Computing (WiCom ’07), pp. 1470–1474, Shanghai, China,
September 2007.
[17] X. Hong, M. Gerla, G. Pei, and C. C. Chiang, “A group
mobility model for ad hoc wireless networks,” in Proceedings
of the 2nd ACM International Workshop on Modeling, Analysis
and Simulation of Wireless and Mobile Systems (MSWiM ’99),
pp. 53–60, Seattle, Wash, USA, August 1999.
[18] F. Bai, N. Sadagopan, and A. Helmy, “Important: a frame-
work to systematically analyze the impact of mobility on
performance of routing protocols for ad hoc networks,” in
Proceedings of the 22nd Annual Joint Conference on the IEEE
Computer and Communications Societies (INFOCOM ’03), vol.
2, pp. 825–835, San Francisco, Calif, USA, March 2003.
[19] K. Fall and K. Varadhan, “The NS Manual,” The VINT Project,
UC Berkeley, January 2002.
[20] S. Gowrishankar, T. G. Basavaraju, and S. K. Sarkar, “Effect of
random mobility models pattern in mobile ad hoc networks,”
International Journal of Computer Science and Network Secu-
rity, vol. 7, no. 6, pp. 160–164, 2007.
[21] T. Camp, J. Boleng, and V. Davies, “A survey of mobility mod-
els for ad hoc network research,” Wireless Communications and
Mobile Computing, vol. 2, no. 5, pp. 483–502, 2002.