International Journal of Computer Applications (0975 – 8887)

Volume 54– No.10, September 2012

Performance Evaluation of ODMRP and ADMR

using Different Mobility Models
K. Kavitha K.Selvakumar
Department of Computer Science & Engg Department of Computer Science & Engg
Annamalai University, Annamalai Nagar Annamalai University, Annamalai Nagar
Tamil Nadu Tamil Nadu

ABSTRACT and planes. It is also used in audio and video streaming or

A Mobile ad hoc network (MANET) is a collection of wire- whiteboard which requires the ability to transmit data to a
less mobile nodes forming a temporary network without using group of receivers.
any existing infrastructure. Mobile Ad-Hoc is a dynamic, Multicasting is the transmission of datagrams to a group of
multi-hop and autonomous networks composed by wireless hosts identified by a single destination address and hence is
mobile nodes. MANETS are subject to rapid and unpredicta- intended for group-oriented computing. Multicasting can effi-
ble changes. MANETS are inexpensive wireless communica- ciently support a variety of applications that are characterized
tion network. The mobility model represents the realistic be- by collaborative efforts and data transmission. Multicasting
havior of each mobile node in the MANET. Routing Protocol techniques can be considered as an efficient way to deliver
for Ad-Hoc networks are typically evaluated using simula- packets from the source to any number of client nodes [2].
tions. In this paper, we compare the performance of On- Numerous multicast protocols have been proposed for mul-
Demand Multicast Routing Protocol (ODMRP) with the ticast in MANETs. These include traditional tree-based or
Adaptive demand driven Multicast Routing Protocol (ADMR) mesh-based protocols such as MAODV, ADMR and OD-
under different mobility models such as Random Way Point MRP. Some multicast protocols use an overlay based ap-
Model, Manhattan Model and Random Drunken Model. Using proach such as AM Route and PAST-DM [3]. The routing
these models, performance metrics such as Packet Delivery protocols selected for the present evaluation study include
Ratio, End-to-End Delay and Control Overhead are evaluated. Adaptive Demand-Driven Multicast Routing(ADMR) and On-
ODMRP dynamically builds the route and manages the group Demand Multicast Routing Protocol (ODMRP). We chose to
membership. In ADMR, Multicast routing state is dynamically compare ADMR and ODMRP because they have been well-
established and maintained only for active groups and only in documented and have been shown to perform well [4].
nodes located between multicast senders and receivers. ADMR
detects the high mobility without the use of GPS or other posi- An ad-hoc routing protocol is a convention, or standard, that
tioning system. The simulation result shows that the through- controls how nodes decide which way to route packets be-
put of ADMR is higher than of ODMR at high mobility and tween computing devices in a mobile ad hoc network [2] [8].
ODMR is high at low mobility. End to end delay and control In this paper, the protocols such as On-Demand Multicast
over head of ADMR is higher than ODMR. Routing Protocol (ODMRP) and Adaptive Driven Multicast
Routing Protocol (ADMR) are proposed. In wireless net-
Keywords working, On-Demand Multicast Routing Protocol is a proto-
MANET, ODMRP, ADMR, Random Way Point model, Ran- col for routing multicast and unicast traffic throughout Ad-hoc
dom Drunken model, Manhattan model. wireless mesh networks. ODMRP creates routes on demand.
This suffers from a route acquisition delay, although it helps
1. INTRODUCTION to reduce network traffic. To reduce the problem of this delay,
Ad hoc networks are dynamically created and maintained by the source will send the first data packet along with the route
the individual nodes comprising the network. They do not discovery packet [3]. In ADMR, source-based forwarding
require a pre-existing architecture for communication purpos- trees are created whenever there is at least one source and one
es and do not rely on any type of wired infrastructure. In an ad receiver in the network. ADMR monitors the traffic pattern of
hoc network all communication occurs through a wireless the multicast source application, and based on that it can de-
media. With current technology and the increasing popularity tect link breaks in the tree, as well as sources that have be-
of notebook computers, interest in ad hoc networks has greatly come inactive and will not be sending any more data [4]. Mo-
peaked. bility models represent the movement of mobile users, and
how their location, velocity and acceleration change over time
MANET is a kind of wireless ad-hoc network, and is a self- [9]. Such models are frequently used for simulation purposes
configuring network of mobile routers (and associated hosts) when new communication or navigation techniques are inves-
connected by wireless links[1]. The topology of mobile ad- tigated [5] [6].In random-based mobility simulation models,
hoc networks is arbitrary. In MANET, the routers are free to the mobile nodes move randomly and freely without re-
move randomly and organize themselves arbitrarily and un- strictions. To be more specific, the destination, speed and
predictably. MANETs require no fixed infrastructure or cen- direction are all chosen randomly and independently of other
tral administration. Mobile nodes in an ad hoc network will nodes. This kind of model has been used in many simulation
work not only as hosts but also as routers, and communicate studies [7]. The Manhattan mobility model uses a grid road
with each other via packet radios. topology. This mobility model was mainly proposed for the
MANET can be used in personal area networking such as movement in urban area, where the streets are in an organized
cellular phone and laptop, emergency operations such as dis- manner. In this mobility model, the mobile nodes move in
aster relief, civilian environments such as meeting rooms and horizontal or vertical direction on an urban map. The Manhat-
sports stadiums, military environments such as soldiers, tanks tan model employs a probabilistic approach in the selection of

35
 International Journal of Computer Applications (0975 – 8887)
Volume 54– No.10, September 2012

nodes movements, since, at each intersection, a node chooses Mobility models can be differentiated according to their spa-
to keep moving in the same direction [1]. In this paper, we are tial and temporal dependencies. Spatial dependency is a
analyzing ODMRP and ADMR protocols by using different measure of how two nodes are dependent in their motion. If
mobility models such as Random Way Point model, Manhat- two nodes are moving in the same direction then they have
tan model and Random drunken mobility model to measure high spatial dependency. Temporal dependency is a measure
the performance metrics such as throughput, delay and control of how the present velocity (magnitude and direction) is relat-
overhead[10][11][12]. ed to previous velocity. Nodes having the same velocity have
high temporal dependency.
2. MULTICAST ROUTING PROTOCOLS
Multicasting is a basic one to many communication way. A Movements of mobile nodes considered in this class are com-
multicast group contains a special host which is responsible pletely uncorrelated. Each mobile node follows an individual
for transmitting data packets to the other hosts in the same independent mobility scenario. Random Waypoint, Manhattan
group [3].In MANET, multicasting can efficiently support a and Random Drunken model belong to this class.
variety of applications that are characterized by close collabo-
rative efforts. Multicast can reduce the communication costs,
3.1. Random Way Point (RWP) Model
link bandwidth consumption, sender and router processing The Random Waypoint Mobility model includes pause times
and delivery delay. In addition, it can provide a simple and between changes in direction and/or speed. A mobile node
robust communication mechanism when the receiver’s indi- begins by staying in one location for a certain period of time
vidual addresses are unknown or changeable. Multicast rout- [5].
ing protocols for ad hoc networks have been proposed in order
to save the network bandwidth and node resource because 200
they are the protocols for powerful communication used in
multi-hop applications, and are more efficient than the ap- 150
proach of sending the same information from the source to
each of the receivers individually. 100
Multicast protocols proposed for MANET can be broadly 50
classified into two categories, namely tree based multicast and
mesh-based multicast. A tree based multicast routing protocol 0
establishes and maintains either a shared multicast routing tree 0 50 100 150 200
or multiple source-based multicast routing trees to deliver data
packets from sources to receivers of a group. In an ad hoc
wireless network, nodes may move freely within the field. For Fig 1: Traveling pattern of mobile node using Random
a pair of nodes to communicate route must be formed between Waypoint Mobility Model
intermediate nodes. For this type of network, it is very im-
portant to model nodes, positions and movement, as transmit- Once the time expires, each node chooses a random destina-
ting range is generally small when compared to the size of the tion in the simulation area and moves towards it with a ran-
field. Multicast plays an important role in ad hoc networks. dom velocity. The mobile node then travel towards the newly
Various multicast protocols have been newly proposed to chosen destination at the selected speed. Upon arrival, the
perform multicasting in Ad hoc Networks such as ODMRP, mobile node pauses for a specific time period before starting
Core Assisted Mesh Protocol (CAMP), Ad Hoc Multicast the process again.
Routing (AMRoute). Figure 1 shows an example traveling pattern of a mobile node
using Random Waypoint Mobility Model starting at a ran-
3. MOBILITY MODELS domly chosen position, the speed of the mobile node in the
Dynamic topology changes in wireless multi-hop network will figure is uniformly chosen between 0 and 10 m/s. In most of
cause lower network connectivity and/or lower network per- the performance study that use the Random Waypoint Mobili-
formance. To capture the nature of mobility of nodes in a ty Model, the mobile nodes are initially distributed randomly
mobile ad hoc network (MANET), different mobility models around the simulation area. When the simulation starts, each
have been proposed. The mobility models used in simulations mobile node randomly selects one location in the simulation
can be roughly divided into two categories: independent entity field as the destination.
models and group-based models. In the independent entity
models, the movement of each node is modeled independently The mobile nodes then travel towards this destination with
of any other nodes in the simulation. In the group mobility constant velocity chosen uniformly and randomly from [0,
models, there is some relationship among the nodes and their Vmax], where the parameter Vmax is the maximum allowable
movements throughout the cells or field. In order to thorough- velocity for every mobile node. The velocity and direction of a
ly simulate a new protocol for an ad hoc network, it is im- node are chosen independently of other nodes. Upon
perative to use a mobility model that accurately represents the reaching the destination, the node stops for a duration
mobile nodes that will eventually utilize the given protocol. defined by the 'pause time' parameter T e. If T e =0, this
Only in this scenario it is possible to determine whether or not leads to continuous mobility. After this duration, it again
the proposed protocol will be useful when being implemented chooses another random destination in the simulation field
mobility model should attempt to mimic the movements of and moves towards it. The whole process is repeated again
real mobile nodes. Changes in speed and direction must occur and again until the simulation ends.
in a reasonable manner. We choose models from different In the Random Waypoint model, V max and Tpause are the two
classes of motion, including random, path-based, and group key parameters that determine the mobility behavior of
based movements. nodes. If the V max is small and the pause time Tpause is long, the
topology of Ad Hoc network becomes relatively stable. On
the other hand, if the node moves fast (i.e., Vmax is large)

36
 International Journal of Computer Applications (0975 – 8887)
Volume 54– No.10, September 2012

and the pause time Te is small, the topology is expected to be

highly dynamic1. Varying these two parameters, especially the
500
Vmax parameter, the Random Waypoint model can generate
various mobility scenarios with different levels of nodal 400
speed.
This mobility model shows that there is high variability in
average mobile node neighbor percentage when using Ran-
300
dom Waypoint model. A neighbor of a mobile node is a node
that is within the mobile node’s transmission range; the aver- 200
age mobile node neighbor percentage is the cumulative per-
centage of total mobile nodes that are given mobile nodes
neighbor. There is a complex relationship between node speed 100
and pause time in the Random Waypoint Mobility model.

3.2. Manhattan Model 0

We introduce the Manhattan model to emulate the movement 0 100 200 300 400 500
pattern of mobile nodes on streets defined by maps. The Man-
hattan map used in our study is shown in Figure2. Maps are
used in this model. The map is composed of a number of hori-
Fig 3: Random Drunken Mobility Model
zontal and vertical streets. Each street has two lanes for each
direction. The mobile node is allowed to move along the grid 4. ON-DEMAND MULTICAST ROUTING
of horizontal and vertical streets on the map [6]. At an inter-
section of a horizontal and a vertical street, the mobile node PROTOCOL (ODMRP)
can turn left, right or go straight. This choice is probabilistic. ODMRP is a multicast routing protocol that uses the concept
The velocity of a mobile node at a time slot is dependent on of ‘forwarding node’ to do the multicasting. It finds some
its velocity at the previous time slot. The Manhattan mobility nodes to be ‘forwarding node’ in the whole network, and only
model is also expected to have high spatial dependence and these nodes will forward multicast messages. The source on-
high temporal dependence. demand establishes the routes by broadcasting the Join Data
message with Time To Live (TTL). This message is periodi-
cally generated to refresh both the membership and routes.
Every intermediate node will add the upstream node’s ID in
its own routing table upon receiving this message. The mes-
sage will be forwarded until it reaches a group member. The
group member then creates a Join Table message and broad-
casts this message to its neighbors.
Every neighbor node will know itself that is on the path be-
tween the source and the group member if the next hop ID in
one of the entries of the Join Table message meets its own ID.
This neighbor node then establishes itself as a forwarding
node. It sets the Forwarding Group Flag on. Then it builds its
own Join Table message based on routing table and propa-
gates it on until the message reaches the source via the short-
est path as shown in figure 4. The mesh of forwarding nodes
Fig 2: Manhattan model is established in this way. This forwarding group supports the
shortest paths between any member pairs. The source can send
3.3. Random Drunken Mobility Model data to all the group members with the help of the mesh. Only
In the Random Drunken mobility model, each node is as- the forwarding nodes will forward the multicast data. It is a
signed a random position within a field. When the node is soft state protocol and there is no need for the group members
next considered for movement, the mobility model checks all to send explicit messages to leave the group. Members can
the possible directions in which the node can move to ensure stop working at any time, and this change can be detected by
that it stays within the field boundaries. The node then moves the periodic refreshment. If nodes in the network have access
in the direction randomly chosen from the set of possible di- to geographical information through equipments like GPS,
rections. Each node moves by one unit distance in that direc- ODMRP then can adapt to node movements by utilizing mo-
tion during the mobility interval. Here the pause time is set to bility prediction. With the mobility prediction method, the
zero seconds. The random drunken model periodically moves protocol becomes more resilient to topology changes.
to a position chosen randomly from its immediate neighboring
positions. The frequency of the change in node position is Mobile nodes forward non-duplicated data packets if they are
based on a parameter specified in the configuration file. Here forwarding nodes. Since all forwarding nodes relay data, re-
each node moves by one unit distance in that direction during dundant paths (when they exist) are available for data packets
the mobility interval. This movement pattern is also known as delivery when the primary path is disconnected. ODMRP also
random walk mobility model. operates as an efficient unicast routing protocol, and doesn’t
need support from another underlying unicast routing proto-
col.

37
 International Journal of Computer Applications (0975 – 8887)
Volume 54– No.10, September 2012

This reply may take one of two forms. If the next scheduled
network flood of an existing multicast data packet is to occur
soon, S may choose to advance the time for this network flood
and use it as the reply for the Multicast Solicitation from R. The other
form that this reply may take is for S to send an ADMR keep-
alive packet unicast to R, following the path taken by R's Mul-
ticast Solicitation packet, as it traveled toward S as shown in
figure 5.

6. PERFORMANCE METRICS
The most important of routing performance metrics are Packet
delivery ratio, End to end delay and Control overhead.

6.1. Packet delivery Ratio (PDR):

Fig 4: Creation of mesh in ODMRP This is the ratio of total number of packets successfully re-
ceived by the destination nodes to the number of packets sent
5. ADAPTIVE DEMAND DRIVEN MULTICAST by the source nodes throughout the simulation.
ROUTING PROTOCOL (ADMR)
ADMR is a multicast routing protocol that cooperates to es-
tablish and maintain forwarding state in the network to allow
multicast communication. ADMR adaptively monitors the 6.2. End to End Delay:
correct operation of the multicast forwarding state and incre- Delay is the time between when a message (CBR data packet)
mentally repairs it when one or more receivers are forwarding was sent and when it was received.
nodes and it becomes disconnected from the sender. ADMR
supports receivers to receive multicast packets sent by any
sender, as well as the newer source specific multicast service 6.3. Control Over Head:
model in which receivers may join a multicast group for only Control overhead is the ratio of the number of control bytes
specific senders. Each multicast packet is dynamically for- transmitted per data byte delivered.
warded from source node S along the shortest delay path
through the tree to the receiver of the multicast group G. Each
multicast packet originated by node S contains a small ADMR
header, including a number of fields used by the protocol in
forwarding the packet and in maintaining the multicast distri- 7. SIMULATION RESULTS AND PER-
bution tree for node S and group G. The sequence number in FORAMANCE COMPARISION
the ADMR header uniquely identifies the packet and is gener- We use the simulation model based on Ns2. In our Simulation
ated as a count of all ADMR packets flooded in any way that the simulated traffic is Constant Bit Rate (CBR) traffic. For
originated from S. The hop count is initialized by S to 0 and is each simulation, we use 100 nodes, randomly placed over a
incremented by each node forwarding the packet. square field whose length and width is 1000 meters. For the
Manhattan model, nodes may only be placed on straight line.
The ADMR header also includes the inter-packet time at
To generate multicast traffic, we use four multicast groups,
which a new packet is expected from the sender S.If the appli-
each consisting of 6 receivers. Each multicast source uses a
cation layer at node S originates no new multicast packets for
Constant Bit Rate (CBR) flow, transmitting a 100 byte packet
G within some multiple of current inter-packet time, the rout-
every 250 milliseconds. Nodes communicate using IEEE
ing layer at S begins originating “keep-alive” packets for G.
802.11 for the MAC protocol, with free space radio signal
The keep-alive is used to maintain the existing forwarding
propagation. We run each simulation for 250 seconds and we
state for the multicast distribution tree for S and G. Absence
average the results of 5 simulations for each data point.
of data packets and keep–alive is an indication of forwarding
tree disconnection. When any source S for multicast group G 8. PERFORMANCE BASED ON MOBIL-
receives the Multicast Solicitation packet (or the single source, in
the case of a source-specific multicast group join), the source re- ITY MODELS
plies to the Multicast Solicitation to advertise to R its existence as a In ODMRP and ADMR, three mobility models namely Ran-
sender for the group. dom way point, Manhattan model and Random drunken mod-
el have been compared based on their performance metrics
such as throughput, end to end delay and control overhead.

8.1. Packet Delivery Ratio

The packet delivery ratio is high in ADMR When compared to
ODMRP in Random way point model (figure 6), Random
drunken model (figure 7), Manhattan Model (figure 8).

Fig 5: Source S1 and S2 respond to receiver R’s Multicast

Solicitation

38
 International Journal of Computer Applications (0975 – 8887)
Volume 54– No.10, September 2012

8.1.1. PDR for Random Way Point Model 8.2.1. Delay for Random Way Point Model

40 40

30 30
PDR

20 ADMR 20

Delay
ADMR
10 ODMRP 10
ODMRP
0
0
5 10 15 20 25
5 10 15 20 25
Speed(m/s) Speed(m/s)

Fig 6: Random Way Point Model

Fig 9: Random Way Point Model
8.1.2. PDR for Random Drunken Model
8.2.2. Delay for Random Drunken Model
40
50
30
40
PDR

20 30

Delay
ADMR
10 ODMRP 20 ADMR
10 ODMRP
0
5 10 15 20 25 0
5 10 15 20 25
Speed(m/s)
Speed(m/s)
Fig 7: Random Way Point Model
8.1.3. PDR for Manhattan Model Fig 10: Random Drunken Model
25 8.2.3. Delay for Manhattan Model
20
50
15
PDR

40
10 ADMR
30
Delay

5 ODMRP
20 ADMR
0 10 ODMRP
5 10 15 20 25
0
Speed(m/s) 5 10 15 20 25
Fig 8: Manhattan Model Speed(m/s)

8.2. End To End Delay Fig 11: Manhattan Model

The Delay is high in ADMR When compared to ODMRP in
all the three mobility models (Figure 9, 10, 11). It is higher for 8.3. Control Overhead
Manhattan model and Random drunken model when com- The Control messages are high in ADMR when compared to
pared with Random way point model. ODMRP in all the three mobility models (Figure 12, 13,
14).Flooding has no control packets and remains relatively
constant and does not increase with mobility.

39
 International Journal of Computer Applications (0975 – 8887)
Volume 54– No.10, September 2012

8.3.1. Control Overhead for Random Way Point ODMRP. At low mobility, throughput is high in ODMRP. It
is observed that delay and control overhead increases for
Model ADMR as compared to ODMRP under different mobility
50 models. Under Random way point model, ODMRP performs
Control overhead

better compared with Manhattan, Random drunken mobility

40 models. In Future these protocols can be compared with some
30 other mobility models and their performance can be analyzed.

20 ADMR 10. REFERENCES

10 [1] B. Malarkodi, P. Gopal, B. Venkataramani, ”Perfor-
ODMRP mance Evaluation of Ad hoc Networks with Different
0 Multicast Routing Protocols and Mobility Models”, In-
5 10 15 20 25 ternational Conference on Advances in Recent Technol-
ogies in communication and Computing, pp.81-84, 2009.
Speed(m/s)
[2] Nurul I. Sarkar and Wilford G. Lol, “A study of manet
routing protocols: Joint node density, packet length and
Fig 12: Random Way Point Model mobility”, in Computers and Communications (ISCC),
2010 IEEE Symposium on, June 2010, pp. 515–520.
8.3.2. Control Overhead For Random Drunken
Model [3] Yudhvir Singh, Yogesh Chaba, Monika Jain, Prabha
Rani, “Performance Evaluation of On-Demand Multi-
casting Routing Protocols in Mobile Ad hoc Networks”,
50 in mobile ad hoc networks.
Control Overhead

40
[4] J. Jetcheva and D. B. Johnson, "Adaptive Demand Driv-
30 en Multicast Routing in Multi-Hop Wireless Ad Hoc
20 Networks”, In ACM MobiHoc,October 2001.
ADMR
10 [5] A. Madani, N. Moussa, “Self-Organized Behaviour
ODMRP Based Mobility Models for Ad Hoc Networks”, Journal
0 of Theoretical and applied Information Technology 2012.
5 10 15 20 25
[6] Niclolas Cooper and Natarajan Meghanathan, “Impact of
Speed(m/s) Mobility Models on Multi-path Routing in Mobile Ad
hoc Networks”, IJCNC Vol2, No1, 2010.
Fig 13: Random Drunken Model [7] Geetha jayakumar, Gopinath Ganapathi, ”Reference
point group mobility and random way point models in
8.3.3. Control Overhead For Manhattan Model performance evaluation of MANET routing protocols”,
Hindwi publication corporation, Journal of Computer
50 systems, Networks, Communication Vol.2008 (2008),
Control Overhead

Article ID 860364.
40
[8] Xu-Zhen, “Bandwidth-Satisfied Multi-path Multicast
30 Routing in Wireless Mesh Network”, ICCP 2010 Pro-
20 ADMR ceedings.
10 ODMRP [9] N. Aschenbruck, E. Gerhands-Padilla, P. Martini, ”A
0 Survey on mobility models for Performance analysis in
Tactical Mobile networks,” Journal of Telecommunica-
5 10 15 20 25 tion and Information Technology, Vol.2 pp.54- 61, 2008.
Speed(m/s) [10] S. Lee, W. Su, J. Hsu, M. Gerla, and R. Bagrodia, "A
Performance Comparison Study of Ad Hoc Wireless
Multicast Protocols”. In IEEE INFOCOM, 2000.
Fig 14: Manhattan Model
[11] J.Kumar and R. Rajesh, “Performance analysis of manet
9. CONCLUSION routing protocols in different mobility models,” IJCSNS
In this paper, ODMRP and ADMR protocols are analyzed International Journal of Computer Science and Network
using mobility models such as Random Way Point model, Security, vol. 9, no. 2, pp. 22–29, Feb.2009.
Manhattan model and Random drunken mobility model to
measure the performance metrics such as throughput, delay [12] Narendra Singh Yadav and R.P. Yadav, “The Effects of
and control overhead. The connectivity of the mobile nodes, Speed on the Performance of Routing Protocols in Mo-
route setup and repair time are the major factors that affect the bile Ad-hoc Networks”, Int. Journal of Electronics, Cir-
performance. The simulation result shows that ADMR is able cuits and Systems, Vol. 1, No. 2, pp 79-84 (2009).
to maintain good throughput at high mobility as compared to

40