Professional Documents
Culture Documents
by
Md Maruf Ilahi
Supervised by
Dr Sahar Al-Sudani
August, 2011
By
Md Maruf Ilahi
A dissertation submitted to
City of London College
in partial fulfilment of the requirements of Degree of
Masters in Computing (Computer Networks)
Awarded by University of Wales
August, 2011
Author Declaration
I hereby confirm that this submission is my own work and that, to the best of
my understanding and credence, it contains no content previously published
or written by any another person (except where explicitly defined in the
references), nor material which to a considerable extent has been submitted
for the award of any other degree or diploma of a university or other
institution of higher education.
_________________________
Md Maruf Ilahi
To my Mother Rawson
Ara Begum for her
inspiration and faith
upon me
Md Maruf Ilahi
Contents
4
List of Figures.......................................................................................... 8
Abstract................................................................................................. 10
Acknowledgement.................................................................................. 11
List of Abbreviation................................................................................. 12
I.
Introduction..................................................................................... 13
1.1
1.2
1.3
Project outline............................................................................... 15
Summery............................................................................................... 16
II.
2.1
2.1.1
2.1.1.1
2.1.1.2
2.1.1.3
2.1.2
2.1.2.1
2.1.2.2
2.1.2.3
2.1.3
Hybrid Protocols........................................................................21
2.1.3.1
2.1.4
Summery............................................................................................... 25
III.
3.1
Routing in MANET.........................................................................26
3.1.1
3.1.1.1
3.1.1.2
Advantages of AODV...............................................................31
3.1.2
3.1.2.1
3.1.2.2
Advantages to OLSR...............................................................33
3.1.3
3.1.3.1
3.1.3.2
Summary............................................................................................... 35
IV.
Research Methodology..................................................................36
4.1
4.2
4.3
Summary............................................................................................... 38
V.
5.1
Performance Metrics......................................................................39
5.1.1
Throughput............................................................................... 39
5.1.2
Network Load............................................................................ 39
5.1.3
Delay....................................................................................... 40
5.2
Simulation Environment.................................................................40
5.2.1
5.2.2
Performance Parameter..............................................................43
5.3
List of Scenarios........................................................................... 44
Summery............................................................................................... 45
VI.
6.1
6.1.1
6.1.2
6.1.3
6.2
6.2.1
6.2.2
6.2.3
6.3
6.3.1
6.3.2
6.3.3
6.4
6.4.1
6.4.2
6.4.3
6.5
Summary............................................................................................... 61
V.
Conclusion............................................................................................ 62
Future Work........................................................................................... 63
REFERENCES....................................................................................... 64
APPENDIX A......................................................................................... 68
Additional Screenshots for chapter 5.........................................................68
List of Figures
FIGURE 1: PROJECT OUTLINE
16
18
28
29
30
33
34
35
40
41
42
FIGURE 12: AODV (15 NODE) RESULTS (AVERAGE DELAY, LOAD AND THROUGHPUT)
47
FIGURE 13: DSR (15 NODE) RESULTS (AVERAGE DELAY, LOAD AND THROUGHPUT)
48
FIGURE 14: OLSR (15 NODE) RESULTS (AVERAGE LOAD, DELAY AND
49
THROUGHPUT)
FIGURE 15: THROUGHPUT ANALYSIS OF AODV, DSR AND OLSR WITH 15 NODES 50
FIGURE 16: THROUGHPUT ANALYSIS OF AODV, DSR AND OLSR WITH 15 NODES 51
FIGURE 17: NETWORK LOAD ANALYSIS OF AODV, DSR AND OLSR WITH 15 NODES
52
FIGURE 18: AODV (30 NODE) RESULTS (DELAY, LOAD AND THROUGHPUT)
53
FIGURE 19: DSR (30 NODE) RESULTS (DELAY, LOAD AND THROUGHPUT)
54
FIGURE 20: OLSR (30 NODE) RESULTS (DELAY, LOAD AND THROUGHPUT)
55
FIGURE 21: THROUGHPUT ANALYSIS OF AODV, DSR AND OLSR WITH 30 NODES 57
FIGURE 22: DELAY ANALYSIS OF AODV, DSR AND OLSR WITH 30 NODES
57
FIGURE 23: NETWORK LOAD ANALYSIS OF AODV, DSR AND OLSR WITH 30 NODES
58
FIGURE 24: OVERALL SCENARIO OF THROUGHPUT (MANET PERFORMANCE
59
COMPARISON)
68
69
70
71
72
Abstract
This research is aim to analyse the performance of routing in Mobile Ad hoc Network, a
network consists of individual nodes connecting with each other creating a infrastructure
less network. Routing in MANET is the most challenging process as constant topology
changes occur during the transmission. A number of protocols are designed to cope with
MANET routing issue.
In MANET each node acts as a host and router while communicating with each other,
routing is one of the core concept in computer networking, so various routing algorithm
inherited by different MANET protocols. In this Research three protocols were evaluated
using Discrete Event Simulator OPNET 16.0. These protocols, (AODV, DSR and OLSR)
use different routing mechanism which is supported by OPNET. A number of scenarios
have been carried out to measure the performance of these protocols; throughput, delay
and load were used as a performance metrics. Two different size of network were
implemented in OPNET to compare the performance.
The results shows that OLSR performs better in both scenario with high traffic load, on the
hand AODV performance became the second best protocol and DSR in most case became
the least performing protocols.
Detail of this performance evaluation is carried out in the chapters.
10
Acknowledgements
I would like to take the opportunity to give gracious appreciation to those people have given
me enormous support and encouragement during this process.
Firstly, I would like to thank my Supervisor Dr Sahar Al-Sudani for constant supervision,
comments and guidance from the start till the end of this Research. Without her contribution
this research would have impossible to achieve to this extent.
I also like to thank those people whose publication and materials I have used to carry out
my research. I would like thank all my colleagues for their support and encouragement
while choosing my topic for the research.
At last I would like thank my Parent specially my Mother for their love, care, inspiration and
trust upon me which is greatly appreciated.
Md Maruf Ilahi
August, 2011
11
List of Abbreviation
AODV
CBR
OPNET
DSDV
DSR
GSR
MAC
MANET
MPR
OLSR
AODV-BR
TCP
RERR
RREP
RREQ
SARP
SP
TORA
ZRP
NS2/3
BATMAN
12
ONE
I. Introduction
Technology changes considerably throughout the last ten centuries, in the tenth or twelve
century, people have remarkable advance in field of science and medical innovation. The
twenty first century became the century of Computer Science, Artificial Intelligence,
Robotics and Space Science.
One of the core computer system concepts is computer networking. This is the age
computer network and internet. The internet most commonly a collection various networks
collaborating and connecting to create a world wide area network also called the
Information Super Highway.
Computer Network consists of two fundamental network concepts, wired network and
wireless network.
Wired Network is a type of network where wire cables are used to design and implement in
certain network. A wired network consists of node, server and link component.
On the other hand wireless network is consists of node, server and frequency link
component.
Two type of wireless network can be categorised, Infrastructure Network and Infrastructure
less network, also called Mobile Ad-hoc network or MANET.
The prime objective of this research to carry out a simulation based performance evaluation
MANET routing protocols. Three protocols were chosen to carry out the research. More
detail will be available in the later chapters.
13
1.1
This thesis is aim to get in depth understanding about the performance of routing in Mobile
Ad-hoc Network and to design, simulate such network with specific hardware and protocols.
MANET has a number of routing protocols and there a number of protocols are on their way
or under development. Each and every protocol has its own characteristics and
performance level, each protocol has diverse mechanism can be suited in different MANET
environment. In this thesis one of the key objectives is to investigate various routing
algorithms and identify the best performing protocols.
1.2
MANET routing is one of most popular subject area for researcher to identify the key
routing protocols to be adopted in different scenario, a number of publications and research
were carried out and a wide variety of key elements is pointed out to provide an idea that
which protocol will be best suited in different environment.
MANET is a type of network where topology changes randomly and no infrastructure is
available for the network to establish an active connection the routing is a vital issue. Most
of the research carried out to determine the performance and security of these protocols
and in numerous researches shows that MANET protocol performance not constant, it is
more ever variable due high mobility, network load and network size. Also the simulation
play vital role in the research as different result shows that simulation environment generate
different outcome as it is need to be configure properly to achieve high level of simulation
efficiency.
The main objective of this research is to identify key protocols which can be used to
determine the performance. So the selection criteria will be one of the most important
factors behind a successful simulation and comparative analysis.
The research questions involves a number of factors as follow,
14
1.3
Project outline
15
Summery
In this chapter an overview of research topic were discussed, aims and objectives and
detail project outline were shown is figure 1. In the next chapter an overview of MANET and
routing protocols will be provided.
16
TWO
II. Overview & Literature Research
Computer Network was always the key element of todays technological advancement. In
previous chapter a brief introduction and detail project outline were discussed. This chapter
is aim to provide an initial of overview of Mobile Ad-hoc network, which is the topic area for
this research and also a literature analysis will be carried out containing similar previous
research materials and outcomes of those researches.
17
Mobile Ad-hoc Network has a number of protocols for difference types of MANET. These
protocols are classified as Reactive, Proactive and Hybrid 1.
II.1.1
Proactive routing protocols use similar approaches as used in some wired network
infrastructure. Proactive routing maintains a number of route table also called vector to
store route information which needs to be updated in regular interval. Every node has the
following node information in the network or subnet along with the node range. But when
topology changes in a network table driven protocols acts differently, but some exceptions
can be found where routing table constantly updated by different protocols. As a results
these protocols needs to maintain few different routing table for each nodes in the network
suggests these protocols are not designed for large MANET as they have to keep
information about every nodes in the network. (Ade & Tijare, 2009)
Network topology changes is one of the key issue in proactive routing as these protocols
needs to maintain up to date topology information as the network topology change
happens, update need to be propagated to the entire network to notify the changes occurs.
Most of the proactive routing protocols use the some mechanisms which are derived from
wired network backbone. So a number of modification and improvements has been carried
out to cope with the dynamic nature of mobile ad hoc network. One of the key
characteristics of MANET proactive protocols is to maintain route update whether any traffic
exist or not in the network, thats make the network load high which can be seen as a
disadvantages of proactive protocols. Some popular proactive protocols are Optimised Link
State Routing, Destination Sequence Distance Vector and Global Source Routing which are
described below. (Liu and Kaiser, 2005)*
get bigger size which uses high amount of bandwidth to transmit the update messages.
(Ablohasan, Wysocki & Dutckiewicz, 2003)
II.1.2
Unlike proactive protocols reactive protocols not always try to search for routes and does
not broadcast control messages in regular interval, reactive protocols search for new routes
as required basis, it also called on demand as the route discovery based on demand. A
route discovery operation carried out when requested by the source node and the operation
terminates as new route discovered or unavailability of routes in the network. Sometimes
an active node connection failure occurred due to mobility of nodes in the network, so route
maintenance operation is the key in reactive protocols, one of the main advantages of
reactive protocols that they transmit less control message comparing to proactive protocols.
But in reactive approach source node often experienced long transmission delay for on
demand route discovery. Two of main reactive protocols are Temporally-Ordered Routing
Algorithm, Ad-hoc On Demand Distance Vector and Dynamic Source Routing. (Liu & Kaiser,
2005)
20
on the node. The protocol is based two algorithm route request approach is determines
route discovery mechanism and route caching determines route maintenance mechanism.
When a node needs to send packet to a destination initially it will look at the route cache for
available route to destination, if these route already expires it will initiate route discovery
operation broadcasting a route request message to its entire neighbour. This Route
Request (RREQ) message consists of source node address, destination address and an
unique identification number. All nodes which receive this RREQ check whether they have
the route in their cache if not they store this route record into their cache and forward the
message on the outgoing link. A node only will send a RREQ if the route is unknown and no
other route request were broadcast to discover this route which can be identified with the
unique number to eliminate duplicate messages. (Royer & Toh, 1999)
II.1.3
Hybrid Protocols
21
II.1.4
Related Work
A number research were conducted by fellow researchers and students in the field of
Wireless Ad-hoc routing performance and a few improvement were proposed to make the
routing more efficient and secure in MANET. MANET routing protocols is one of the most
popular subject area for postgraduate Computer Science and Telecommunication students
in recent years. MANET routing protocols are different with difference performance level
when size and environment varies and these protocols need to be dynamic in nature to
adopt the environment which the protocols deployed.
A number of studies are outlined as follow in a form of table to provide greater
understanding about MANET Protocol performance and reliability.
Table 2.1: A list of research carried out about MANET protocol performance.
Author
Implementation
S. Ade &
Protocols
Summary of Results
P. Tijare, 2010
node mobility.
M. Rahman et al,
AODV, DSR,
2009
W. Lol, 2008
OPNET Modeller
OLSR, TORA
A. Zaballos, et al,
AODV, DSR,
2003
OLSR, TORA
Johnson,
Maltz,
Real life
DSR
Result varies
Low routing overhead in random
Implementation
mobility
.
A group of researcher (Johnson, Maltz, & Broch, 2001) carried out experimental simulation
involving Dynamic Source Routing protocol to measure the performance and security with a
real life case study designed to measure performance in rapid mobility environment. DSR
performs well in multi-hop wireless ad hoc network, as the simulation 6 results with a real life
ad hoc network of car driving and routing within the network, with low routing overhead and
its ability to accurately transmit all originated data packets, even with continues, random
motion of all devices in the network DSR provides excellent results. ( Johnson, Maltz, & Broch,
2001)
Recent study in 2010 ( Ade & Tijare, 2010) shows the simulation outcomes concur with
predictable findings based on theoretical investigation. As expected reactive routing
protocol AODV7 performs better in the view that it is capable to continue the connection in
any interruption while transmitting data packets, which is needed in TCP 8 connection-based
traffic. AODV predictably deliver all packets virtually at low node mobility and not capable to
maintain connection at increased mobility. On the other hand DSR is preferred the best
protocol in all mobility rates and the speed is quite similar to DSDV, But DSDV is costly in
increased mobility compare to DSR. In comparison to on demand (AODV, DSR) with table
driven (DSDV) measuring the metrics like end to end delay and data dropped rate DSDV
performs better in any network with small number of nodes while DSR/AODV performance
is relatively superior in a network with large number of nodes. ( Ade & Tijare, 2010)
Another comparative study (Rahman, Islam & Talevski, 2009) shows similar results as packet
dropping rate for DSR is less than DSDV and AODV indicating its highest efficiency. The
authors simulated only three protocols (AODV, DSR and DSDV) and concluded the best
performing protocols in high mobility is AODV and DSR while DSDV provide better
performance in low mobility with fewer nodes in the network. High mobility responses with
link failure and data overhead while updating all nodes with available routing information
which can be found much more in DSDV than AODV and DSR. A general conclusion can
be reached while observing the application oriented metrics like packet delivery fraction and
delay, Performance of DSR is much better than DSDV and AODV and also the routing load
is low in DSR than the other two protocols. ( Rahman, Islam & Talevski, 2009)
In 1999 a review was published by Royer and Toh ( 1999) concerning various routing
protocols and their performance and uses in Mobile ad hoc network. The review provides
information about protocols and their features. Its also included algorithm used by those
protocols and advantages and disadvantages of each protocols. The authors stated in the
paper as each and every protocols has their own characteristics and features for different
scenario and each protocol can perform well in their most suited scenario, The author
conclusion are as follow,
Finally, we have identified possible applications and challenges facing ad hoc mobile
wireless networks. While it is not clear that any particular algorithm or class of algorithm is
23
the best for all scenarios, each protocol has definite advantages and disadvantages, and is
well suited for certain situations. (Royer & Toh, 1999)
In a study of MANET protocols in 2003 a group of researcher (Zaballos, Vallejo, Corral,
Abella, 2003) point out few key performance issue of four different routing protocols (AODV,
OLSR, DSR, TORA) At a first glance, the results indicates that proactive protocols initiate a
minor delay in the network, as their route discovery is on demand. However, due to the fact
that they constantly try to explore new routes to all feasible destinations, packet overhead
in routing is high. In contrast, reactive protocols do not retain vacant routes and look for
new routes when necessary. This process increases the delay suffered by packets,
because they are in the buffer queue waiting to be transmitted. Usually they produce fewer
control traffic than the proactive routing protocols.
The study also suggested OLSR9 performance varies in term of high mobility with large
number of nodes interacts in the network. Firstly is it so unlikely that OSLR would be able to
discover the routes quickly, so the coverage time increases. Secondly it continuously
searches the routes for all probable destinations in the network and causes high control
traffic. Finally OLSR is recommended in an environment with low mobility and small number
of nodes to get the best result out of it. But in terms of other protocols such as AODV the
study indicates that these protocols can provide best all around performance and also its
development over DSDV and DSR protocols will make it a highly resourceful and adaptive
protocols to apply in MANET. Similarly DSR performs in high data load scenario but causes
relatively high delay while network size increased. In such environment, route length
become longer and increase the size of packet length constantly which results the uses of
this protocols to be restricted to small and medium wireless network. But in Temporally
Ordered Routing Algorithm (TORA) the main advantage is that it can generate multiple
alternate routes to destination, which is supported in multicast environment. But in some
Wireless ad hoc network cases TORA10 is vulnerable to crumple in the network, which will
results a protocol collapse. Notably the simulation results suggested that TORA
performance is worse because it produce large number of traffic which indicates it is not
one of the recommended protocols where critical level of performance ensures network
efficiency. But the results also suggests that in a network with large number of nodes
involves TORA does not underperform and could minimize network delay. (Zaballos,
Vallejo, Corral, Abella, 2003)
According to a recent study (Lol, 2008) by one of the graduate student from the Auckland
University of Technology New Zealand shows some rather improved results of four MANET
routing protocols (AODV, DSR, OLSR, TORA), the research suggested that in highly
demanding environment with medium number of nodes AODV perform well but when the
mobility get higher this protocol may not able to survive the pressure. Likewise the
simulation results indicate that TORA can be an option to implement in small network when
mobility is high and in medium size network with high mobility and traffic load. (Lol, 2008)
In most network in term of size where mobility is rather low, OLSR adopt really well in most
of the cases, the type of data traffic used is vital for this kind of positive results. The study
(LOL, 2010) also point out that OLSR performs exceptionally well with CBR 11 traffic, which
is the only traffic used by the author to simulate OLSR. Finally the author concluded that for
small network DSR might be the best option, AODV can be the option for medium network
and for large number of node involvement TORA can be the winning protocols, while OLSR
predominantly the best routing protocols for any network in term of size, traffic load and
mobility, but in some cases OLSR underperform due to mobility and traffic load.
So while reviewing the above study (Lol, 2008) results a final solution can be reached that
within a small network with high mobility DSR and OLSR loose performance while TORA
and AODV gain reliability, but it react opposite while number of node decreased. So overall
the best protocols for all size of networks still need to developed or redeveloped.
Summery
In this chapter in depth literature review is provided. A initial overview of MANET and
Protocols are also discussed. In the next chapter detail routing algorithm will be discussed.
Three
III. Routing in Mobile Ad-hoc Network
In the previous chapter an overview about MANET routing provided with an in depth
literature research has been carried out to achieve common understanding about similar
research done by different student and researcher. This chapter focused on more detail
understanding about MANET routing algorithm and a selection protocol mechanism will be
discussed which later will be used as a experimental protocols for this research.
III.1 Routing in MANET
One of the core concepts in networking is routing no matter the network is wireless, wired
or other architecture. A number of protocols were designed for MANET in last decades to
cope with network challenges in MANET and some of them were really popular and
extensively used for MANET. Distance Vector routing and Link state routing are both
dynamic routing widely used in wired network. These algorithm is used to design MANET
routing protocols, this two are described in below in detail with the MANET associated
protocols.
III.1.1
Distance Vector routing algorithm is on e of the core routing algorithm used in both wired
and wireless network based on classical Bellman-Ford routing algorithm. Distance Vector
26
uses a vector to store the routing information for its neighbour. The router passes distance
information with its neighbour in regular interval to update the routing table. These distance
determined by the actual hop number between each node as well as queue length and
delay occurred in the transmission. This algorithm always tries to use the shortest path if
there is a number of routes involves between two nodes. One of the main disadvantages of
this algorithm is the coverage is low which leads to a count to infinity problem. (Liu &
Kaiser, nd). The most well known protocol which uses distance vector algorithm is Ad-hoc
On-demand Distance Vector designed for MANET which is an updated version of DSDV
protocol. (Liu & Kaiser, nd)
Route Request
In AODV, when a host node try to send a message to another node but there is no route
information available which leads to a Route Discovery operation. In Route discovery
operation the host node will broadcast a RREQ message to its neighbours, which includes
source destination address and broadcast ID, a unique identifier. It also contains up to date
sequence number of the routes. As shown in the figure 3 there are five nodes involved in
this example A, B, C, D and E. In the figure node a wants to sends a message to node C
but does not have the route to node C, so node A generate a RREQ and broadcast to its
neighbour node B and D. Node B and D receive the RREQ and updates its routing table if
the sequence number new or greater than the number it has have in its table and the route
27
is known. From the figure it can be said that the RREQ message contains both source and
destination number, also have a ID and lifespan which is 3. So both neighbours node B and
D gets the RREQ but only B has the route to the destination to C, so B will generate a
Route Reply message for A includes the route hop count and will create a RREQ for node
to have a possible route to its destination if these route not even known by node B. On
other hand the other neighbour node D which is not in range to the destination can try to
get an alternative through broadcasting a RREQ to node E. In the figure 4 a detail Route
Reply operation is discussed.
Route Reply
From the above figure 3 it can be said that node A finds an active route to node C through
node B. When node B sends a RREP to node A its also sends RREQ to node C to get the
route. In AODV all the route information is kept on the route table in a vector. So when node
C receives a route request from node B as seen in figure 4 below, it will update its own
route table with path to node A via node B. It also generates a route reply for node B to let
the node know that a route path is discovered to node A which will enable an active
connection. From the figure 4 it can be seen the RREP message format which includes
source node, destination node, hope count and ID. On the other hand node D finds out that
the route to node C is not reachable from this node so node D just rebroadcast the
28
message to its neighbour node E. This whole operation establishes an active connection
between node A and C. (C. Liu & J. Kaiser)
Route Error
In AODV, node uses Hello message to get touch with its neighbours. In regular interval
each node broadcast a hello message to notify the neighbour its existence. If there is a
problem in connection or disconnection occurs the nodes generate a Route Error message
and will broadcast it to let the neighbour know about the connection problem. From the
figure 5 it can be seen that node B want to send a message to node C but there is no active
connection, and gets a Route Error (RERR) message from node C which indicates a
possible connection failure between two nodes. This connection failure can be repaired
using another techniques called local repair, which is a distinctive feature of AODV
algorithm. (C. Liu & J. Kaiser)
29
Local Repair
Local repair is a type of routing operation when link with other nodes broken and the host
node need to repair it if the target node nearest hope as specified in the route table. The
host node tries to repair it generating RREQ message with an increased sequence number
and broadcast the message to the destination host.
measured because the repair operation must not extent through the whole network. During
the process of repair after sending RREQ to the target the host node waits until it gets the
RREP message from the target node if no respond received with certain amount of time it
will update its routing table with the value invalid or it will update the hop count after
receiving the reply. Sometime the hop count is greater than its already been in the routing
table than a new RERR with that hop count will be broadcasted. This link repair process
always considered as proactive repairing where target node route is unavailable before
data can be sent locally.
30
Multicast Routing
Multicast routing also supported in AODV which makes it a versatile protocol for MANET.
Multicast routing concept includes the routing where the IP addresses a sequence numbers
are kept. It can also record the hop count and leader IP addresses. RREQ messages are
used to link with multicast groups in the network and RREP messages are used to make
respond by the nodes. In multicast routing a source node can receive a number of RREP
from different nodes in the network but the source node will choose the shortest path route
to the destination via using a number of node within the route. In a multicast group tree an
MACT (Multicast ACTvation) message used to send activation link into the branch.
Sometimes there are no RREP received by source which anticipates that there are no
active nodes connected in the multicast tree. A multicast RREP consists of IP address,
group leader and hop count of the neighbouring nodes. When group node get RREP
messages from two different group leaders which involves a tree connection between two
trees. Within connected in a multicast group a member can leave the group tree anytime
but it has to generate a MACT o for the branch it belongs, otherwise it has to continue
serving as a group member.
31
messages, which are responsible for the route maintenance, are also limited so that they
do not create unnecessary overhead in the network. The AODV protocol is a loop free and
avoids the counting to infinity problem, which were typical to the classical distance vector
routing protocols, by the use of the sequence numbers. (Huhtonen, 2004)
III.1.2
Link State routing uses Dijkstras algorithm, which uses different metrics to calculate the
route information. Notification messages were sent by each router to the other routers in
the network with updated link and topology information. When a router receives a
notification messages from a router it will update routes with the new topology information.
This is how a router will be keep updated about the whole network scenario. Different
metrics are used to compare the route information for example, number of hops, link speed
and traffic congestion. Link State routing uses Shortest Path technique to transmit data
messages. Optimize Link State Routing (OLSR) one of the popular Link State algorithm
based protocol.
OLSR constantly maintain the link information of the whole network and update the routing
table, this can cause a large number of overhead traffic throughout the network. This
reduce the control messages OLSR use Multi Path Relay technique, in MPR within the
network a small number of MPR nodes are chosen to maintain link traffic in the network.
That means it create a MPR network within a set proximity to reach the nearest proximity.
Application used in OLSR is designed to reduce long delay during transmission of packets.
In the small network proximity OLSR tries to obtain low number of MPR sets as possible
which will maintain low route traffic. But one disadvantage of this protocol is it tries to keep
updated routing table for the whole network, which is both similar for small or large network,
but when the number of nodes increases the overhead of control messages goes up.
(Ade & Tijare, 2010)
32
33
III.1.3
Source routing uses transmission data to maintain route information, each node uses the
original message header to send the route data to other nodes. A node receiving any route
request will add its own source address to the responding route reply and forward the
packet to the originating node. So in that way the source node gets up to date route
information for its routing table. In this algorithm also helps to reduce the loop by identifying
the packet header. DSR is one of the well known Source Routing protocol.
34
Summary
This chapter concentrates on detail routing algorithm of MANET protocols. Some figure will
help to understand the routing approaches. The next chapter will elaborates research
methodology for this research.
35
Four
IV. Research Methodology
In the previous chapter a detail routing techniques for MANET were discussed and some
examples were presented. In this chapter the methodology which will be used to carry out
the research will be the main focus. A number of simulator and its main characteristics will
be observed. This chapter is divided into 3 sections, in the first section research
methodology will be discussed, a number of simulation models will be shown in section 2
and in the third section a justification of using such methodology will be discussed.
36
using different modules. Other notable simulators are GloMoSim also a DES Open Source,
QualNet, OmNet++. Most of these simulators are developed with C++ or C so C++ compiler
is necessary runs the modules integrated in these highly effective programs.
A number of well known simulation engine is listed in the table below
OPNET
GloMoSim
Commercial
Educational
Simulation
/Open Source
Support
Type
Commercial
Yes
DES, Object
Oriented
Library Based
etc
Wireless Network
Open Source
Yes
Protocol Supported
Parallel
QualNet
OmNet++
Commercial
Open Source
No
Yes
Distributed and
Wired, Wireless,
Parallel
WLAN etc
DES Modular
Wireless Network
Component
NS-3
Open Source
YES
Based
DES, Object
Multicast routing,
Driven
TCP, MANET,
Wireless etc
37
(Optimized Network Engineering Tools) is another Discrete Event Simulator with object
oriented functionalities. Between NS2 or NS3 and OPNET consideration can made
because these two simulators was commonly used to simulate MANET protocols in recent
years. Especially for this project OPNET Modeller is selected for its distinct behavioural
characteristics. Justifications to choose OPNET Modeller are as follow,
Summary
In this chapter the concept of computer research methodology were discussed, then a
number of simulator engine is listed in a table. These simulators later described and a
simulation engine is chosen to carry out the research. OPNET Modeller 16.0 is chosen to
use to simulate the protocols AODV, DSR and OLSR. In the next chapter using OPNET a
different network models and simulation will be discussed.
38
Five
V.
In previous chapter the research methodology is discussed with detail of simulation tool is
presented. In this chapter firstly the performance metrics will be discussed and secondly
network model will be presented. The performance metrics are in one part and the network
model with detail scenario and configuration will be discussed in another part.
5.1
Performance Metrics
The performance metrics is used to determine the performance of a routing protocol. These
metrics measure a network performance using standard units. To measure a protocol
performance in a network a number of metrics are used, in this research the selected
metrics are Average Throughput, Average Network Load, End to end delay.
5.1.1
Throughput
Throughput determines the average rate of packets arrived over a transmission channel
and the unit its measured in bit per second (bits/second). Its also measures the efficiency
and effectiveness of the protocol performance and determines the network performance
from one node to destination. Its also analyzed the quality of route and the capacity of
routing algorithm over network load.
5.1.2
Network Load
Network load is the routing loads which can be define as the number of routing traffic is
being transmitted over the number of data packet transmitted from a source to its
destination. It also determines the numbers of overhead packet are being transmitted
through the network. In other words the traffic overhead is the number of control message
transmitted to destination. Network load is measured with bits/sec unit.
39
5.1.3
Delay
The delay can be measured in different ways; firstly the duration a packets spends at the
queue during transmission. It also determines the buffer time and propagation time delay.
Delay can be categorised as the network efficiency while using maximum resources by a
network protocol.
5.2
Simulation Environment
The simulation is carried out using Discrete Event Simulation (DES) tools called OPNET
Modeller 16.0, which already discussed in the research methodology chapter (Chapter 4).
The simulation is divided into 6 scenario based on network sizes on three MANET protocols
AODV, OLSR and DSR, the first three scenario consisting 15 nodes with random mobility
and high network load, and the remaining three scenario consisting 30 nodes with the same
configuration.
5.2.1
Two types network model shown in figure 6 and 7 consisting 15 and 30 nodes respectively.
More detail is provided later parts in this chapter.
40
As shown in the figure 9 and 10 examples network models were presented with 15 and 30
wireless LAN nodes, Application Configuration, Profile Configuration, Mobility Configuration
and a Wireless LAN server. In each scenario these objects were configured with the
protocols used in the scenario.
Simulation model consists of 5 different types of objects as detailed below,
Node Model: Node model consists of 15 Wireless LAN advanced node which are
configured with each protocols used in the experiments. A node is a simple wireless
device which also acts as a router in MANET. A node model architecture is shown
on figure 11 below,
41
Wireless LAN Server: Wireless LAN server the only server which is configured
with each protocol used and with FTP data packets for transmission. This server
serves all the nodes in the network involving a number of configurations. A
screenshot is provided in the appendix section for detail.
42
5.2.2
Performance Parameter
A number of parameter for network simulation model is listed below shown in different
tables,
Table 5.1: Common Parameter used in simulation
Parameter
Area
Network size
Data Rate
Mobility Model
Network Interface Type
Link Layer Type
Large packet processing
Wireless LAN
Data type
Simulation Time
Mobility speed
Value
1000 x 1000 sq meter
15 & 30
11 Mbps
Random Waypoint
Wireless Physical Layer
Data Link layer
Fragment
IEEE 802.11e capable
FTP
1800 Seconds
10 m/s
Values
5
10
(uniform) (1,11)
10
2
0.04 sec
35
2
3
Enabled
Parameter
DSR
Value
43
Parameter
Value
Willingness
default
64
2.0
16
TC interval (sec)
5.0
10
6.0
0.5
15.0
0.03
30.0
50
5.3
0.25
2
0.5
List of Scenarios
A number of scenarios were created and simulated for this project, but only a list of 6
successful scenarios is presented in the paper, each scenario consists of few different
experiments in term of protocols versus performance metric. These scenarios are shown in
a table below,
01
02
03
04
05
Description
In this scenario the protocol employed was AODV with 15
nodes and 1800 seconds of simulation time. The maximum
range was 1000 x 1000 meters with random mobility
configured. The ftp traffic load was set in High Load. Three
performance metrics Throughput, Load and Delay were set
for the results to compare the performance.
This scenario involving DSR having 15 nodes with 1800 seconds of
simulation time. The range is the same as AODV as the load is high.
Performance is measured with WLAN Throughput, Load and Delay.
The only difference in scenario 3 is the protocol used is OLSR, so the
network is configured as with OLSR parameter. Earlier these parameters
were discussed in a form of table in this chapter.
Scenario 4 involves AODV again but with changed configuration, where the
network size increased but the simulation time is less now which is 30 nodes
and 900 seconds respectively. The performance parameter is same as
previous scenario. In this scenario more advanced network configuration is
used to cope with the bigger size of the network.
06
In this scenario the default protocol used for routing is OLSR, so the
configuration needs to change to cope with OLSR protocols.
Summery
In this chapter detail about the simulation environment is provided in form of detail writing,
figures and tables. This chapter is one of the key chapters in this project as its the start
point of the main practical experiments which will follow in the next chapter.
Six
45
46
Figure 12: AODV (15 node) results (Average Delay, Load and Throughput)
As shown in the above graph the AODV protocol get good results as the delay started from
just below 0.015 seconds decrease gradually and till the end of the simulation, the average
delay is just 0.02 seconds. On the other hand network Load started from 600,000 bits/sec
but decline steadily to a rate of 100,000 which is the average Load AODV protocol. And the
throughput starts from high at over 600,000 and finished at near 50,000 bits/sec which can
be said as constant 70,000 in average. Detail data were presented in the table below,
Start at
Finish at
Results (Average)
Throughput (bits/sec)
620,000
50,000
70,000
Delay (sec)
0.013
0.01
0.002
Load (bits/sec)
600,000
40,000
50,000
47
The next graph shows the results of DSR protocols throughput, Load and
Delay in figure consisting of three different graphs. Scenario 02 involving
DSR protocol with 15 nodes in the network, the Simulation Time is set to
be 1800 sec with high load traffic.
Figure 13: DSR (15 node) results (Average Delay, Load and Throughput)
In the graph above its can be said for when using DSR in the network results indicate high
rate of throughput about 600,00 bits/sec at the start of the simulation but decreases as time
passes and reach the average at about 70,000 bits/sec. On the other hand Network Load is
quite similar as throughput as seen in the graph but same result happens as it decreases
average level of 50,000 bits/sec at the end of the simulation. At the start the Delay was at
0.026 sec but get average about 0.08 sec at the end of the simulation. DSR gets low
number of delay as time passes at simulation. But results indicate AODV gets lesser rate of
delay than DSR.
48
Scenario 02
Started at
Throughput (bits/sec)
Delay (sec)
Load (bits/sec)
>600,000
0.026
600,000
Finished at
Results (Average)
>50,000
0.05
40,000
60,000
0.08
60,000
Figure 14: OLSR (15 node) results (Average Load, Delay and Throughput)
From the graph above it can be determined that OLSR get better throughput results than
both AODV and DSR, starts from below 600,000 bits/sec but keep good rate at about
200,000 at the end of the simulation, gets about 250,000 bits/sec in average. Similar
results can be seen for delay and where OLSR achieve an average of below 0.001 sec
delay and 50,000 bits/sec Load comparing to other two the results represents that OLSR
performs better. Detail of the results presented in table below,
49
Scenario 03
Start at
Finish at
Results (Average)
Throughput (bits/sec)
Delay (sec)
Load (bits/sec)
550,000
0.015
450,000
150,000
0.001
70,000
200,000
<0.001
35,000
Figure 15: Throughput Analysis of AODV, DSR and OLSR with 15 nodes
From the above graph it can be seen that three coloured lines represents throughput of
three protocols, blue colour symbolize AODV, red stands for DSR and OLSR can be seen in
colour green. The average throughput of OLSR is far better that the other two protocol as
OLSR uses link state algorithm, while AODV with distance vector and DSR with a
50
combination of link state and source routing algorithm. OLSR starts with rather low rate of
throughput at the start comparing to other two but keep a good rate till the end as the other
two AODV and DSR loose the throughput rate gradually at a low rate at the end. OLSR
achieve over 200,000 bits/seconds of throughput comparing 70,000 bits/sec and 60,000
bits/sec for AODV and DSR respectively. A table of chart is included below to make the
comparison more constructive,
From the graph the average throughput for OLSR is 200,000 far better than other two
above, so it can be said that OLSR performance is better for small size network with 15
nodes, high load traffic, random mobility and 1800 seconds of simulation time. More detail
analysis will be available at the end of this chapter.
Figure 16: Throughput Analysis of AODV, DSR and OLSR with 15 nodes
51
Figure 17: Network Load Analysis of AODV, DSR and OLSR with 15 nodes
Scenario 01, 02 & 03 (AODV, DSR & OLSR = 15 nodes & ST = 1800 seconds)
Protocol
Average Throughput
Average Delay (Sec)
AODV
(bits/sec)
<70,000
>0.002
>50,000
DSR
>60,000
<0.008
<60,000
OLSR
>200,000
>0.001
<35,000
From the above comparison table it can be anticipated that OLSR has the highest of
average throughput and the average delay is less than the other two, OLSR also has
lowest number of average Load comparing to other two AODV and DSR. On the other hand
between DSR and AODV, AODV transmitted the most number of successful data packets
through the network as the throughput is a bit if compare with DSR and AODV has low
delay and low amount of network load.
52
Figure 18: AODV (30 node) results (Delay, Load and Throughput)
53
Start at
Finish at
Results (Average)
Throughput (bits/sec)
>1,100,000
<400,000
>580,000
Delay (sec)
<0.013
<0.002
>0.005
Load (bits/sec)
<780,000
>200,000
<420,000
Figure 19: DSR (30 node) results (Delay, Load and Throughput)
In scenario 5 figure 19 indicates more significant changes in DSR with large network size,
the average throughput recorded is just below 270,000 which is much greater than previous
DSR experiments as expected, and also the delay went up a bit caused by network size,
which is 0.017 in average. Average Load collected 220,000 which are quite normal in large
network. In table 6.5 the statistics of DSR scenario 5 experimental shown. The table 6.6
below shows the average throughput, delay and load in a listed way,
54
Start at
Finish at
Results (Average)
Throughput (bits/sec)
<1,200,000
>150,000
<270,000
Delay (sec)
<0.046
<0.009
>0.017
Load (bits/sec)
>1,056,000
<130,000
>220,000
Figure 20: OLSR (30 node) results (Delay, Load and Throughput)
In scenario 6 OLSR protocol with 30 (figure 20) node experiments are done. A number of
performance statistics collected and shown in table 6.5. Average throughput recorded over
one millions packet were successfully transmitted through the network which is quite
remarkable results for OLSR, although the network size gets bigger this time. The Average
55
load was 210,000 quite improved load compared with high amount of throughput. And
0.002 seconds of average delay which is standard better performance in protocol delay.
Table 6.6 Shows the statistics collected from OPNET 16.0 as follow,
Start at
Finish at
Results (Average)
Throughput (bits/sec)
>1,500,000
>850,000
>1,000,000
Delay (sec)
<0.012
<0.001
>0.002
Load (bits/sec)
<950,000
>160,000
<210,000
56
Figure 21: Throughput Analysis of AODV, DSR and OLSR with 30 nodes
Figure 22: Delay Analysis of AODV, DSR and OLSR with 30 nodes
57
Figure 23: Network Load Analysis of AODV, DSR and OLSR with 30 nodes
Scenario 04, 05 & 06 (AODV, DSR & OLSR = 30 nodes & ST = 1800 seconds)
Protocol
Average Throughput
Average Delay (Sec)
AODV
(bits/sec)
>230,000
>0.003
<360,000
DSR
>260,000
<0.010
<260,000
OLSR
>800,000
>0.001
<250,000
58
In the above graph 6 different scenarios is added which makes a comparison graph, from
previous statistics in terms of throughput it can be said that for small network OLSR
performance were best among three protocols, 2 nd best performance shown by AODV, in
increased node OLSR performance get even better with high number of throughput. AODV
has done slightly better than DSR.
So it can be anticipated that in small network environment OLSR will perform better with
high throughput, even better in large number of nodes involves.
59
As per as network delay concern again OLSR gets low delay rates than any other in the
same network, Also AODV gets much less delay than DSR in the network. DSR became
the least performing protocols in terms of delay and throughput.
60
But from figure 26 some improvement in DSR can be seen, when DSR gets less number of
load when increased network environment. But in small network environment DSR still
underperform in term of network Load. On the other hand OLSR performance affected
when it get most number of Load in the comparison scenario with 30 nodes network. Also
AODV underperform in the same large network.
Above all the scenario is clearly judged and a number statistics were provided as evidence
in performance comparison, different size networks were used to compare the same
protocols, a conclusion can be reached as OLSR become the number 1 protocol in both
small and medium sized network. The reason behind OLSR performance is its reactive
protocol which uses Multipath Routing technique. So traffic over can be reduced which can
minimize network load in the network. But in some network where network size get bigger
OLSR can produce problem as the Load is really high as seen in the simulation results.
On the other hand AODV gets good performance review in both small and large network,
distance vector routing algorithm remain one of the most vital protocol suite for routing in
MANET. But in some area AODV get lower results than DSR and OLSR as both of these
protocols can reduce the overhead using different algorithm.
DSR was least performing protocol above all of them in both small and large network
model. But in some cases DSR can reduce Network Load and can perform better.
Summary
In this chapter a number of graphs and statistic charts were presented produce better
performance review. OLSR performance is better as shown from those results. The next
will be the end note of this process where a number of key points will be discussed; also a
future research concept will be shared.
61
Seven
V.
The previous chapter outlines a number of MANET experimental scenarios results which
indicates a performance comparison between three MANET routing protocols (AODV, DSR
and OLSR). In this chapter a summary of the whole project will be included with a future
research proposal will be discussed.
Conclusion
The test results shows for small size network AODV and DSR outperforms compare to
OLSR, the throughput rate is really high of OLSR protocols in small size network of 15
nodes with high ftp traffic.
OLSR achieve a extensive value of performance from this research, OLSR gets better
results in all three performance parameter. Because it is a reactive routing protocol when
using advanced MPR technique to relay the route request, it is found effective in both of the
OLSR scenario in this research. OLSR can identify multiple MPR nodes in a neighbour
network and will use shortest path to the destination choosing from multiple path in the
network.
AODV and DSR also had shown good performance in both small and large network, but
comparing with OLSR these protocols nowhere near to beat the link state algorithm based
protocol. In some cases OLSR performance not up to the standard but in most of the cases
it has great level performance in MANET based network.
AODV and DSR based on Distance vector and Source Routing can perform better in small
network with low load network traffic but in high load traffic they underperform, one of the
reason behind AODV because its proactive nature, periodical HELLO message
transmission which affect on network Load and can produce end to end delay.
Also DSR performance in under the water because DSR uses same kind of technique to
discover route in the network.
62
Future Work
The future research area is based on different AODV OLSR reproduced algorithm, As
AOVD-BR (Ad hoc On Demand Distance Vector Backup Route) is the new breed of
AODV algorithm. This new algorithm has integrated back up routing mechanism which can
be compared with alternative route exist in OLSR.
Other research topic can be under developed BATMAN protocol which will replace OLSR in
near future. BATMAN stands for Better Approach to Mobile Ad-hoc Networking is a Linux
kernel 2.6.38 based which is a part of Lurgo-Mesh Project. It uses an intelligence technique
to discover the route. It is still under development will be available to research later on. A
simulation based on Linux platform can be performed which focuses on the performance
and stability will level in challenging Mobile Ad-hoc Network.
63
REFERENCES
1. Lee, S. and Gerla, M. (2000) "AODV-BR: Backup routing in ad hoc networks."
Proceedings of IEEE WCNC 2000, (September) Chicago, pages 1311-1316.
2. Jacquet, P. Muhlethaler, P. and Qayyum, A. (2002) "Optimized Link State Routing
Protocol", IETF Internet Draft, draft-ietf-manet-olsr-10.txt.
3. Royer E. and Toh. C., (1999) A Review of Current Routing Protocols for Ad-Hoc
Mobile Wireless Networks, IEEE Personal Communications Magazine, April 1999,
pp. 46-55.
4. Das, S. R., Perkins, C. E. and Royer, E. (2000) Performance comparison of two ondemand routing protocols for ad hoc networks. Proceedings of the IEEE InfoCom,
March 2000.
5. Tanenbaum, A. (1998) Computer Networks, 4th edition, Prentice Hall.
6. Perkins, C. and Royer, E. (nd) Ad-hoc on-demand distance vector routing,
Technical report, Sun Micro Systems Laboratories, Advanced Development Group,
USA.
7. Vincent D. Park and M. Scott Corson. Temporally-Ordered Routing Algorithm (TORA)
version 1: Functional specification. Internet-Draft, draft-ietf-manet-tora-spec-00.txt,
July 2001
8. Abolhasan, M., Wysocki, T., and Dutkiewicz, E. (2003) A review of routing protocols
for mobile ad hoc networks. Technical report, Telecommunication and Information
Research Institute, University of Wollongong, Wollongong, NSW 2522; Motorola
Australia Research Centre, 12 Lord St., Botany, NSW 2525, Australia.
9. Broch, J., Maltz, D., Johnson, D., Hu, Y. and Jetcheva, J. (1998) A performance
comparison of multi-hop wireless ad hoc network routing protocols, Proc.
64
MOBICOM, pp - 85-97.
65
20. Webb, J. (2005) Analysis of packet flows in simulated Ad-hoc networks using
standard network tools Masters thesis, University of California Santa Cruz, USA.
[Online].
Available
at:
http://citeseerx.ist.psu.edu/viewdoc/download?
doi=10.1.1.115.7593&rep=rep1&type=pdf
21. Rangarajan, H. (2006) Robust loop-free on-demand routing in Ad- hoc networks
PhD Dissertation, (June) University of California, USA. [Online]. Available at:
http://citeseerx.ist.psu.edu/viewdoc/download?
doi=10.1.1.61.8458&rep=rep1&type=pdf
22. Mosko, M. (2004) Routing in Mobile Ad-Hoc Networks, PhD dissertation, (June)
University
of
California,
Santa
Cruz,
USA.
[Online].
Available
at:
http://citeseerx.ist.psu.edu/viewdoc/download?
doi=10.1.1.3.5144&rep=rep1&type=pdf
23. OPNET Inc, (2006), OPNET University Program,, [Online]. Available at:
http://www.opnet.com/university_program/research_with_opnet/ [Accessed:
June 01, 2010]
24. CCNA., (nd) Routing protocols and concepts, CCNA exploration companion guide.
Introduction to dynamic routing protocols. Chapter three pages 148 to 177.
25. Gowrishankar, S., Basavaraju, T., Singh, M. and Sarkar, S. (nd) Scenario based
Performance Analysis of AODV and OLSR in Mobile Ad hoc Networks, Jadavpur
University, Acharya Institute of Technology India
26. Rahman. M., Islam. M., Talevski. A., (2009) Performance Measurement of Various
Routing Protocols in Ad-hoc Network. Proceeding the International Multi Conference
of Engineers and Computer Scientists IMECS (March 18 20, Vol - I) Hong Kong.
27. Bertocchi, F., Bergamo, P., Mazzini, G. and Zorzi, M. (nd) Performance Comparison
of Routing Protocols for Ad Hoc Networks, DI, University of Ferrara, Italy.
28. Ade, S. and Tijare, P. (2010) Performance Comparison of AODV, DSDV, OLSR and
DSR Routing Protocols in Mobile Ad Hoc Networks. International Journal of
Technology and Knowledge Management, Vol 2 (July - December), pp 545 548
29. Suri, P., Soni, M. and Tomar, P. (2010) Cluster based QoS Routing Protocol for
MANET. International Journal of Computer Theory and Engineering, Vol 2
(October).
66
30. University of Texas. (2010) Format Guideline for Doctoral Dissertation and
Dissertation Abstracts. University of Texas Graduate School (September 2010)
31. OPNET Technology, Inc. (2007) How To: Design Mobile Ad hoc Network and
Protocols. Best Practice Approaches for Accelerating Network R&D, OPNET
Technology Incorporated (January)
32. Hogie, L., Bouvry, P. and Guinand, F. (2006). An Overview of MANET Simulation.
University of Luxemburg.
APPENDIX A
Additional Screenshots for chapter 5
67
68
69
70
71
72