MEE 09: 17
MASTER THESIS REPORT
MSc IN ELECTRICAL ENGINEERING
WITH EMPHASIS ON TELECOMMUNICATION
TRASMISSION CONTROL PROTOCOL (TCP)
PERFORMANCE EVALUATION IN MANET
BLEKINGE INSTITUTE OF TECHNOLOGY
MARCH 2009
Author:
Muhammad Ijaz
m_ijazf@yahoo.com
Supervisor & Examiner:
Adrian Popescu
adrian.popescu@bth.se
1
ACKNOWLEDGEMENT
In the name of greatest All mighty ALLAH who has always bless me with potential
knowledge and success.
I am thankful to my supervisor Adrain Popescu, my friends who help me during my hard
times when I need their assistance during simulation.
I am especially thankful to my Parents, Uncles and Grandmother, who had always provided
me the courage, strength, best wishes, moral and financial support during my whole career.
I also have best regards for BTH faculty including Mikeal Åsman and Lena Magnusson who
had been helpful throughout my maters degree.
2
Abstract
Mobile Ad hoc network routing protocols have been divided in several different categories
such as Reactive and Proactive Routing Protocol. The performances of these categories are
evaluated in different scenario with TCP variants. We present a comprehensive TCP
performance evaluation study to understand the nature of the TCP performance in different
scenarios with variable amount of payload and number of nodes. The traffic consists of three
different packet sizes i.e. 512, 1000, 1500 bytes each. Three different routing protocols
(AODV, DSR and TORA) are to be evaluated with three different TCP variants (Tahoe, Reno
and New Reno) in three different scenarios having 3, 5 and 8 nodes. The performances
parameters on the basis of which routing protocols are to be graded are mainly throughput,
congestion window and delay. Conclusions are drawn based on the simulation results and the
comparisons between them have been elaborated.
3
LIST OF OBSERVATIONS
ABR
Associativity -Based Routing
ACK
Acknowledgement
ALP
Adaptive Link-State Protocol
AODV
Ad hoc On Demand Distance Vector
BER
Bit Error Rate
CLR
Clear
CWND
Congestion Window
DAG
Directed Acyclic Graph
DREAM
Distance Routing Effect Algorithm for Mobility
DSDV
Destination Sequence Distance Vector
DSN
Distance Sequence numbers
DSR
Dynamic Source Routing
FORP
Flow Oriented Routing Protocol
FSR
Fisheye State Routing
GLS
Grid Location Service
GPS
Global Positioning System
GPSR
Greedy Perimeter Stateless Routing
GSR
Global State Routing
IEEE
Institute of Electronics and Electrical Engineers
IP
Internet Protocol
LAN
Local Area Network
LANMAR
Landmark Ad Hoc Routing
LAR
Location-Aided Routing
LMR
Lightweight Mobile Routing
LRR
Least Resistance Routing
MANET
Mobile Ad hoc Network
MAC
Medium Access Control
MDSR
Multipath Dynamic Source Routing
MSS
Maximum Segment Size
OLSR
Optimized Link State Routing
OPNET
Optimized Network Evaluation Tool
PHY
Physical Layer
RABR
Route-Lifetime Assessment Based Routing
4
RERRs
Route Errors
RIP
Routing Internet Protocol
QRY
Query
RREQs
Route Requests
RREP
Route Reply
RTO
Retransmission Time Out
RDMAR
Relative Distance Micro-discovery Ad Hoc Routing
ROAM
Routing On demand Acyclic Multipath
RWND
Advertize Window
SSA
Signal Stability-Based Adaptive
STAR
Source Tree Adaptive Routing
SWND
Send Window
TCP
Transmission Control Protocol
TORA
Temporally Ordered Routing Algorithm
TTL
Time-To-Live
UDP
User Datagram Protocol
WRP
Wireless Routing Protocol
ZHLS
Zone-Based Hierarchical Link State
5
LIST OF FIGURES
FIGURE
Figure 2.1
Figure 2.2
Figure 3.1
Figure 3.2
Figure 3.3
DESCRIPTION
Slow Start and Congestion Avoidance Mechanism
Hidden and Exposed Node Terminal Problem
Route discovery for target node
Maintenance for Error Route
Route Discoveries in TORA – QRY Message
Figure 3.4
Route Discoveries in TORA – Update Message
26
Figure 4.1
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Research Methodology
Simulation Model
MANET eight node scenarios
Simulation Flow
30
32
33
34
36
37
38
39
40
Throughput comparison in three nodes scenario
Throughput comparison in five nodes scenario
Throughput comparison in eight nodes scenario
Delay Comparisons in Different Scenario
Congestion Window dynamics
PAGE
15
18
23
24
26
LIST OF TABLES
TABLE
Table 5.1
DESCRIPTION
Detail of TCP Variants and Routing Protocol for Three Nodes
PAGE
35
Table 5.2
Detail of TCP Variants and Routing Protocol for Five
36
Table 5.3
Detail of TCP Variants and Routing Protocol for Eight Nodes
37
6
LIST OF CONTENTS
CHAPTER 1
1.1
1.2
1.3
1.4
1.5
Introduction
Background
Aims/ Objectives
Research Questions
Thesis Structure
Expected Contribution
11
11
11
11
11
CHAPTER 2
2.1
2.1.1
2.1.2
2.2
2.2.1
2.2.2
2.2.2.1
2.2.2.2
2.2.2.3
2.3
2.3.1
2.3.2
2.3.3
2.3.4
2.3.5
2.3.6
2.3.7
TCP and Ad hoc Networks
12
Types of Wireless Networks
Infrastructure Networks
Ad hoc Networks
TCP working mechanism and Variants of TCP
Working Mechanism of TCP
Various Flavor of TCP
Tahoe
Reno
New Reno
Challenges and Limitation of TCP in Wireless Network
Dynamic Topology/Mobility
Multi Hopping
Differentiating Transmission and Congestion Losses
High Bit Errors
Channel Contention
Hidden and exposed terminal problem
Out of Order Packet
13
13
13
14
14
15
15
16
16
16
16
17
17
17
17
18
19
CHAPTER 3
3.1
3.2
3.2.1
3.2.2
3.2.3
Ad hoc Network Routing Protocols
20
Proactive routing protocols
Reactive routing protocols
Dynamic Source Routing (DSR)
Ad hoc on Demand Distance Vector (AODV)
Temporarily Order Routing Algorithm (TORA)
22
22
23
24
25
CHAPTER 4
4.1
4.2
4.3
4.3.1
4.3.2
4.3.3
4.3.4
4.4
Research Methodology
28
Research Methodology
Author Approach
Research Design
Problem Selection / Identification
Literature Review for TCP and MANET
Building Simulation Environment
Simulation Results
Simulation Tools
29
29
29
29
29
30
30
30
9
7
CHAPTER 5
5.1
5.1.1
5.1.2
5.1.3
5.1.3.1
5.1.3.2
5.1.3.3
5.1.3.4
5.1.3.5
5.2
5.3
5.4
Simulation and Empirical Study
31
Simulation Environment
Model
OPNET
Network Entities and Functions
Application Definition
Profile Configuration
Mobility Configuration
Server
Mobile Nodes
Simulation Configuration
Simulation Flow
Scenarios
32
32
32
32
33
33
33
33
33
33
33
34
CHAPTER 6
6.1
6.2
Discussion/ Results
42
Conclusions
Future Work
43
43
Appendix
44
Reference
60
8
Chapter 1
INTRODUCTION
9
1.1
Background
Internet evolution thrives with time and development in the diverse technology
improvements. Communication devices and medium have significant impact on overall
internet. Numerous networks mixtures are present at the internet and even within these
networks different preposition and parametric approaches for implementations are deployed
subject organization needs. Overall internet traffic is somehow dependent upon underlying
networks and vice versa. To cope with multiple simultaneous data networks internet requires
and efficient algorithmic techniques to support data transmission successfully. Transmission
Control Protocol/Internet Protocol (TCP/IP) has been very effective in data transmission
delivery and have also developed variants to possess the possibility to increase performance
and multiple packet loss recovery. An important aspect form [1] defines the data flow method
that applications are responsible for the selection of the protocols and IP plays a central role
all TCP and user datagram protocol (UDP) for data transmission.
TCP is reliable and connection oriented protocol developed in 1981. It’s based on simple
sliding window flow control, during the early stages congestion collapses occurred because of
lack of congestion control mechanism. With the advent of Jacobson congestion algorithms for
TCP as a remedy, TCP is updated to its new version known as TCP Tahoe. Now currently
TCP Reno is widely used in Internet. TCP Tahoe congestion Control includes slow start,
congestion avoidance and fast retransmission. While in addition to these three algorithms,
TCP Reno also adds the fast recovery algorithm. TCP is design for wired network, but with
the technology emerging towards wireless medium, the need to implement TCP is of great
important but it faces many problems especially in an ad hoc networks. Mobile ad hoc
network (MANET) is a famous ad hoc network can be utilized well for emergency situation
and military applications. TCP has poor performance in MANTE due to dynamic topology,
shared medium, high error ratio; channel connotation and multi hop architecture.
As an evident, ad hoc networks are growing dynamically and its true development fact lies in
the problems associated with the seamless internet access and connectivity. With evolution in
the hardware devices and software that offers wireless access to the internet by means of an
intermediate connection through access point ad hoc tends to improve and increase gradually.
Mobile ad hoc network is sub field with enormous commercial, industrial and military growth
and realistic practical approach for internet access. Most important things to know about the
technology is the ability of autonomously establishing and managing the network [2] while
the an effective military consensus applications, data exchange based on mobility networks
for commercial purpose i.e. business organization or it could be better utilized during
emergency within and out of remote areas. This IP based dynamic wireless networks works
effectively, reliable and robust data transmissions.
Problems are associated with in the MANET performance, an evaluation and optimization
techniques are necessary to opt and adhere for the better execution of the transmission
medium. Pledged data delivery is TCP utmost drawback in wireless networks but possible
solutions are available to recognize the data transmission effects. To measure the performance
of different TCP variants, simulation study has been conducted in practice. MANET utilizes
TCP and UDP for data transmission and our study focus on different variants of the TCP i.e.
particularly Tahoe, Reno and New Reno explicitly using AODV, DSR and TORA protocols
in focus.
10
1.2
Aims/ Objectives
Thesis aims to understand the performance of the TCP variants in MANET and its behavior
with respect to different protocols. It focuses following issues in the thesis:
Detail literature study for “state of the art” technology in TCP and MANET.
Performance comparison in delay and throughput by increasing number of nodes.
Performance comparison between the different protocols used in MANET through
congestion window, delay and throughput.
Suggest better protocol and TCP variant for MANET.
1.3
Research Questions
Q1:
Q2:
Q3:
Q4:
What are different TCP variants and its need?
What are important routing protocols used in MANET?
What is the impact on TCP performance during different protocol selection/
implementation in MANET?
Which factors influence the TCP performance and why?
1.4
Thesis Structure
Chapter 2 defines wireless networks and an extension to ad hoc networks and its working, it
also include the detail study about the TCP flavors and its limitations.
Chapter 3 is associated with the study of MANET routing protocols specific to ad hoc
networks, their types, working and algorithms.
Chapter 4 depicts the overall project flow explicitly starting form problem definition and
following it towards solution, implementation, simulation and results hierarchy.
Chapter 5 contains simulation environment description, list of figures and assumptions.
Chapter 6 holds the brief discussion about the simulation results and future work.
1.5
Expected Contribution
The thesis will address the scientific society about the TCP and MANET implementations in
reality. It will contribute through using multiple different numbers of nodes to describe the
behavior of the TCP performance over different variants; it should reveal the idea to
developing commercial and non commercial business project for MANET deployments and
the results in reality should have utmost significance.
11
Chapter 2
TCP AND AD HOC
NETWORKS
12
In this chapter first to discuss the taxonomy of wireless networks such is infrastructure
networks and ad hoc networks or infrastructure-less networks. Then before to discuss the
different challenges transmission control protocol (TCP) is facing in wireless networks, it is
better to give a look to the TCP working mechanism and its different variants.
2.1. Types of Wireless Networks
One of the unique features of wireless networks is compare to wire network is that data is
transmitted from one point to another through wireless links i.e. there is no need of wired link
between the two nodes for transmission. They just need to be in the transmission range of
each other. Wireless networks or divided into two categories. Infrastructure wireless network
and infrastructure less or ad hoc wireless network.
2.1.1. Infrastructure Networks
Infrastructure network have fixed network topology. Wireless nodes connect through the
fixed point known as base station or access point. In most cases the access point or base
station or connected to the main network through wired link. The base station, or access point,
is one of the important elements in such types of networks. All of the wireless connections
must pass from the base station. Whenever a node is in the range of several base stations then
it connect to any one of them on the bases of some criteria [3].
2.1.2. Ad hoc Networks
Ad hoc networks also called infrastructure less networks are complex distributed systems
consist of wireless links between the nodes and each node also works as a router to forwards
the data on behalf of other nodes. The nodes are free to join or left the network without any
restriction. Thus the networks have no permanent infrastructure. In ad hoc networks the nodes
can be stationary or mobile. Therefore one can say that ad hoc networks basically have two
forms, one is static ad hoc networks (SANET) and the other one is called mobile ad hoc
networks (MANET). From the introduction of new technologies such as IEEE 802.11[4] the
commercial implementation of ad hoc network becomes possible [5].
One of the good features of such networks is the flexibility and can be deployed very easily.
Thus it is suitable for the emergency situation. But on the other side it is also very difficult to
handle the operation of ad hoc networks. Each node is responsible to handle its operation
13
independently. Topology changes are very frequent and thus there will be need of an efficient
routing protocol, whose construction is a complex task. TCP performances are also very poor
in mobile ad hoc network. In coming sections we are discussing the TCP working mechanism
and challenges for TCP in ad hoc networks in more detail.
2.2 TCP working mechanism and Variants of TCP
Transmission Control Protocol (TCP) [6] is a Transport Layer Protocol and originally
designed for wired network in 1981. The basic responsibility of TCP is to provide reliable
transfer of data between the nodes i.e. to ensure that the data is reached the destination
correctly without any loss or damage. The data is transmitted in the form of continuous stream
of octets. The mechanism is adopted to assign a sequence number to each octet of data and
receiver respond with positive acknowledgement to ensure that the data is received correctly.
Another aspect of TCP is the tree way handshakes mechanism to establish a connection
between the end points (nodes). In case if there are two or more TCP connection between the
end points for this TCP uses the port assignment mechanism to differentiate each connection
from others.
2.2.1 Working Mechanism of TCP
The congestion control algorithm employed by TCP is window based. It uses three types of
windows called congestion window (cwnd), advertize window (rwnd) and send window
(swnd). Congestion windows shows the total amount of data the sender is allowed to output to
the network without any Acknowledgement (ACK). While on the other hand the advertise
window indicates the amount of data the receiver is ready to accept. It is equal to the available
buffer size on the receiver. Now the size of the flight window or send window is set to the
minimum of the congestion window and advertize window.
Basically the congestion control mechanism has two phases the slow start and congestion
avoidance phase. When a connection is established the initial size of congestion window is set
to One MSS (Maximum Segment Size). Receiving data from the sender the receiver ACK the
reception of data to the sender, the receiver is actually indicating the sequence number of next
expected data segment in ACK, from this the sender conclude that all the data segments
which have sequence number less than the indicated one are delivered correctly. While in case
of packet loss, out-of-order packets arrive at the receiver. Then the receiver sends a duplicate
14
ACK to the sender in response of each out-of-order packet arrival. When sender receives three
duplicates ACK, it is concluded that the packet is lost and retransmission of data packet take
place.
Slow Start threshold
Slow Start threshold
Time out Occured
24
22
20
Congestion Window
18
16
14
Congestion Avoidance
Phase
12
Slow Start
10
8
6
4
2
0
0
1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16
Number of Acknowledgement
Figure 2.1 Slow Start and Congestion Avoidance Mechanism [10]
If sender receives non-duplicate ACK Then the congestion window is incremented. In slow
start phase exponential incrimination occurs until it reaches to the slow start threshold. While
in congestion avoidance phase the cwnd is incremented by one packet. This window growth is
limited to a maximum window size. This whole process is depicted in Fig. 2.1.
2.2.2. Various Flavor of TCP
After the introduction of first version of TCP several different flavors exist, here we are
discussing the most famous implementation of TCP called Tahoe, Reno, New Reno and
Vegas.
2.2.2.1. Tahoe
In the first version of TCP there was no congestion control mechanism. So after observing the
congestion collapses 1988 Jacobson introduced several Congestion Control algorithms [7] and
this version is called TCP-Tahoe. The congestion control algorithms introduced in this version
are:
15
a) Slow start
b) Congestion Avoidance
c) Fast Retransmit
2.2.2.2. Reno
The fast retransmit phase was first introduced in TCP-Tahoe followed by Slow Start. But
TCP-Reno [8] also added the algorithm of Fast Recovery, so that Fast Recovery dictates the
sender to perform congestion avoidance directly after fast retransmission rather than
immediately reducing the data flow using slow start mechanism.
2.2.2.3. New Reno
TCP Reno recovers only one lost packet during the recovery process. So TCP-New Reno [9]
is just adding the capability to TCP Reno to deals with multiple packets losses to recovery in a
single transmission window.
2.3. Challenges and Limitation of TCP in Wireless Network
As compare to wired network wireless network have some special characteristic such as
Mobility, share medium and signal fading. All these factors are responsible to create different
types of challenges that restrict the range, data rate and reliability of the wireless transmission.
These challenges are as follow.
2.3.1. Dynamic Topology/Mobility
In contrast to wired network, in wireless ad hoc networks the devices are free to move
which leads to frequent topology changes. Then two types of problem occurrence are
possible [11].
(a) Path loss
(b) Network Partition
Path loss leads to the path re-computation at sender side and during this phase there will
be no transmission which ultimately causes throughput degradation. Another possibility
during this phase is that if path re-computation take more time then may be retransmission
time out (RTO) occurs. Thus there will be data retransmission, RTO is increased
exponentially and TCP enter to slow start phase. This affect will be more serious in high
16
mobility environment. In case of network partition the sender and receiver remains at
different network and all the packets will be dropped. In such situation multiple
consecutive transmission of the same segment are possible, while the receiver will be
disconnected and it is called serial time out.
2.3.2. Multi-hoping
In Ad hoc Network Environment every node is also supporting the responsibility of router to
forwards packets on behalf of other nodes. Thus longer flow, in since of hops, having longer
round trip time and higher packet dropping probability and high fluctuating end-to-end
throughput as compare to flow having small number of hops.
2.3.3.Differentiating Transmission and Congestion Losses
Since wireless network have open medium, signal passing through the medium they have to
face different blockage such as building, due to these blockages the transmission signals are
reflected, diffracted and scattered and thus causes packets loss. TCP assume that this loss is
occurred due to congestion and it activate its congestion control and result in congestion
window reduction, which leads to low throughput and low utilization of available bandwidth.
So TCP has no such mechanism to differentiate between congestion losses and transmission
losses.
2.3.4 High Bit Errors
In wired network the Bit Error Rate (BER) is ranging from 10-6 to 10-8, while in wireless
network this range is from 10-3 to 10-1. Since due to high BER the packet losses are very
frequent and TCP reaction to these losses reduces the congestion window. Thus leads to nonoptimal performance [10].
2.3.5 Channel Contention
Another reason of TCP performance degradation is the channel contention due to increasing
number of nodes. It can occur between different flows passing through the same vicinity or
between different packets in the same flow. In IEEE802.11, when the number of try for
channel access exceeds the predefined limit, then cause to drop the packets and the Medium
Access Control (MAC) Protocol notifies (wrongly) the upper layer that the path is
17
unavailable. In response the upper layer starts the route recovery procedure [12] and TCP stop
its transmission and the throughput drop to zero during route recovery process.
This channel contention also leads to unfairness problem. The unfairness also occurs between
the nodes so that each node has or not the equal access to the medium as compare to other
node. If there is no equal access to the medium for each node then the unfairness occurs
between the flows passing from different nodes. This unfairness also happens among the
flows passing from the same path.
2.3.6 Hidden and exposed terminal problem
Due to the share medium and multi hoping capability the nodes facing the hidden and exposed
node problem. Fig: 2.2 representing the hidden and exposed terminal problem of IEEE
802.11 standard. The circles show the transmission range of A and B, where C is in the
transmission range of both A and B. Let A and B both want to transmit data to C, so there will
be collision at C, because A and B do not know about the transmission of each other due to
hidden node problem.
Figure 2.2: Hidden and Exposed Node Terminal Problem
Now let that there is a transmission between C and B, while at the same time A wants to
transmit data to D, but when A senses the medium, it find that the medium is busy due to C
transmission and thus A stops its transmission. Actually in this situation A transmission for D
will not going to collide with C transmission, This problem exist in IEEE802.11 standard and
known as exposed terminal problem.
18
2.3.7 Out of Order Packet
When a receiver receives out of order packets, the receiver transmits duplicate
acknowledgement, after receiving three duplicate ACK the sender retransmit the packets and
congestion control is activated. But the problem is that congestion control is activated
wrongly most of the time, because out-of-order packet occurrence take place due to different
reasons such as multipath routing protocol and rout failure and not only due to congestion.
19
Chapter 3
Ad hoc Network Routing
Protocols
20
In this chapter routing protocols are described. In general route means the way and protocol is
the set of rules through which two or more devices (computers, mobile nodes) are
communicating with each others. Routes are multi hop in ad hoc networks because the
propagation range (250 meters in an open field) of wireless radio is limited. Nodes travel
freely and randomly in the network and routes are often find connection or disconnection.
Establishing strong routes, maintaining and reconstruction in time are the main task for
routing protocols. All the above responsibilities are performed by the routing protocol, except
generating excessive control message overhead. Data packets send efficiently must be utilized
by control packets and be generated only when needed. Routing protocol efficiency in
bandwidth and energy consumption could be made by reducing the control overhead. [13]
In generally mobile ad hoc wireless networks routing protocols can be classified by their
routing strategy.
Pure distance vector algorithms (e.g., Distributed Bellman Ford, Routing Internet
Protocol (RIP), etc.) do not give a good result in mobile networks because of some
limitation. Then some new protocols were proposed to modify and enhance the
distance vector algorithm. Protocols such as Wireless Routing Protocol (WRP), Least
Resistance Routing (LRR), Destination Sequence Distance Vector (DSDV) routing
protocol, and the protocol by Lin and Liu.
The protocols which are based on link state algorithms. Theses protocols include
Global State Routing (GSR), Landmark Ad Hoc Routing (LANMAR) protocol,
Optimized Link State Routing (OLSR) protocol, Adaptive Link-State Protocol (ALP),
Fisheye State Routing (FSR) protocol, and Source Tree Adaptive Routing (STAR)
protocol.
The third one is on-demand routing protocols which are planned only for ad hoc
network. Route to every destination of the networks on a regular basis is not
maintained by on-demand routing protocols. The source establishes routes on demand.
The source floods a route request packet to construct a route when it needed. The
destination use route selection algorithm and select the best route for which
destination receives request. Then route reply packet is sent to the source through new
best route. There is no requirements of periodic exchange of route tables and control
traffic overhead is greatly reduce by on-demand routing protocols. Several protocols
of this type have been propose d. Ad-Hoc On Demand Distance Vector (AODV)
21
routing, Dynamic Source Routing (DSR), Lightweight Mobile Routing (LMR),
Temporarily Ordered Routing Algorithm (TORA), Route-Lifetime Assessment Based
Routing (RABR), Associativity-Based Routing (ABR), Relative Distance Microdiscovery Ad Hoc Routing (RDMAR) protocol, Signal Stability-Based Adaptive
(SSA) routing, Multipath Dynamic Source Routing (MDSR), and
Routing On
demand Acyclic Multipath (ROAM) algorithm are on demand routing protocol.
The fourth category is GPS (Global Positioning System), in the early stages protocols
was using node location information while building routes have been proposed
recently. Through information node position, routing can require more cost to
exchange location information. GPS routing protocols are Greedy Perimeter Stateless
Routing (GPSR), Distance Routing Effect Algorithm for Mobility (DREAM), Grid
Location Service (GLS), Location-Aided Routing (LAR), Flow Oriented Routing
Protocol (FORP), and Zone-Based Hierarchical Link State (ZHLS).
In addition we can say routing protocols are categorized as reactive, proactive and hybrid
(combination of both is called hybrid). [13]
3.1 Proactive Routing Protocols
This type of protocols has to construct and maintain fresh routing information to all the nodes.
This is free of whether or not the route is needed. For this achievement control message are
transmitted periodically. Proactive routing protocols are not bandwidth well-organize. Even if
there is no data flow, the control message is broadcasted. This type of protocols has some
advantages and disadvantages. One of the main advantages is that nodes can get easily routing
information and easy to set up a session. The disadvantage is:
There is too much data kept by the nodes for route protection and restructure is slow when
there is a failure in exacting link. DSDV is an example of proactive routing protocol.
3.2 Reactive Routing Protocols
Reactive routing protocols are specially planned for ad hoc networks. Permanent routing
information is not kept by these protocols. Routes are built when the source needed. Route
request is sending across the network to achieve this. In reactive routing protocols we describe
in detail DSR (Dynamic Source Routing) protocol, AODV (Ad Hoc On-demand Distance
Vector) protocol, and TORA (Temporally Ordered Routing Algorithm) protocol. [13]
22
3.2.1 Dynamic Source Routing (DSR)
DSR was developed at Carnegie Mellon University. It is simple and efficient reactive routing
protocol which is specially designed for multi-hop ad hoc network of mobile nodes. The
nodes in the network easily join or leave the network without any information. The network
using DSR is not requiring existing network infrastructure or administration. The node
desiring to transmit a packet define route for the packet because it is based on source routing.
DSR works for ad-hoc network of approximately 200 nodes. Each node participating in adhoc network should forward packets and discard the erroneous packets (corrupted). DSR has
two mechanisms: route discovery and route maintenance. [13]
Route discovery
The source starts a route discovery when sending data packet to the destination
but have no routing information. To set up a route, the source floods RREQs
message with a distinctive request ID. When the destination receives this
request message or a node which has destination route information then it
transmits RREP message back to the source with route information. Figure 3.1
shows route discovery of DSR. Node 2 is the initiator and node 9 is the target.
Figure 3.1: Route discovery for target node, [19]
Route Maintenance
In LAN routing the main improvement of DSR is in route maintenance and
monitoring in the attendance of mobility. DSR based on the acknowledgments of
data packets sent to adjacent nodes to monitors the validity of existing routes. This
monitoring is achieved by inactively listening for communication of the adjacent to
the next hop or sitting a bit in a packet to ask for open acknowledgment. The
RERRs packet is sent to the creative sender to raise a new route discovery stage
when a node fails to accept an acknowledgment. Nodes receive a REERs message
remove any route entry (from their route cache) which uses the out of order link.
23
When a node has problem transferring packet during that link then REER message
is propagated. So this selective transmission reduces control overhead (if no packets
pass through a link), it yields a long delay when a packet wants to go through a new
link. [13]. Node 9 cannot be reached by node 6 anymore and a REER is returned to
node 2.
Figure 3.2: Maintenance for Error Route, [19]
DSR main advantages are that it reduce routing overhead and does not need to discover routes
to all the nodes in the network. The disadvantage of DSR is low mobility and static networks.
Its performance is reduced by high mobility.
3.2.2 Ad Hoc On-demand Distance Vector (AODV)
The Ad hoc On-Demand Distance Vector (AODV) is a routing protocol. AODV is designed
for ad hoc mobile networks and of both routing, that is unicast and multicast routing. AODV
establish routes between different nodes as needed by source nodes. AODV maintain these
routes as well as form trees which connect different multicast group members. The group
members compose the trees and the members are connected by the nodes. In an ad hoc
network when two nodes want to make a connection with each other, AODV enable multihop routes within the nodes. Ad hoc On-demand distance vector is free loop. DSN (Distance
Sequence numbers) is used by the AODV to avoid counting to infinity, and this is one of the
most important quality and feature of this algorithm. In a network the requested nodes send
the DSN with other routing information from the source to the destination. It has also the
feature to select optional route which is based on the sequence number. [14]
There are three messages which are defined by AODV. These messages are Route Errors
(RERRs), Route Request (RREQs) and Route Replies (RREPs). [15].
24
For discovering and maintaining routes in the network these three messages are used, by using
UDP packets from source to destination. A node uses its IP address as the source address in
the IP header of a message when it request for a route, and for broadcast 255.255.255.255. In
the ad hoc network the number of hops a particular routing message is determined by the TTL
(Time-To-Live). The RREQ is broadcasted by the requested node when a route is needed to
be created to the destination. When the next hop node received the message a new route is
determined, or itself by the destination [16]. Routes of the PREQ from the originator to all the
stations that receive message are cached in these stations. A RERR message is delivered or
generated when there is a failure in the link. The message has all the important information
about the node which is not reachable because of the failure. The IP address of the nodes is
also mentioned in the message as the next hop for the destination.
AODV is table based. All the information about the routes in the network is stored in this
table. The routing table has the following entries i.e. DSN, flag, next hop, IP address, State,
hop count, the list of precursors, Life time and network interface.
3.2.3 Temporally Ordered Routing Algorithm (TORA)
TORA is a reactive routing algorithm based on the conception of link reversal and used in
MANETs to improve the scalability. Highly dynamic mobile ad hoc networks can be used by
TORA. It is an adaptive routing protocol and used in multi-hop networks. It makes scaled
routes between source and destination and Directed Acyclic Graph (DAG) is used to build in
the destination node. A DAG is accomplished for the network by handing over each node i
with a height metric i.e. hi. Directional from the link i to j means hi > hi . The height of the
node in TORA is defined as quintuple, which contains the unique ID of the node, a
propagation ordering parameter, the logical time of a link failure, a reflection indicator bit and
a unique ID of the node which defines the new reference level. The first two values identify
an offset with respect to the reference level and the last three values jointly represent the
reference level. A data packet goes from up flow to down flow according the height
difference between nodes. TORA has the capacity that many nodes can send packets to a
given destination is provided by DAG. It also guarantees that all routes are loop free. There
are three basic functions in TORA, [17 and 20].
Route Creation
Route maintenance
25
Route erasure
A route creation operation start all the nodes height is set to NULL i.e. (-,-,-,-,i)
and
destination is set to ZERO i.e. (0,0,0,0,dest). The nodes height is flexible according to the
topology. A QRY message containing the destination’s ID is broadcasted by the source. A
non-NULL height node responds with UPD packet containing its own height. With the
reception of UPD packet, a node increment its height with one than that of the UPD generator.
The higher height of a node is measured as up flow and lower height of a node is measured as
down flow. Then the DAG from source to destination is constructed. Route maintenance
operation is a main part of TORA. It has an exclusive quality that control message are
contained into a small set of nods near the happening of topology changes. A node generates a
new reference level and broadcasts the reference to its neighbours when it loses its last down
flow link. Then the links reproduce the topology change and adjust to the new reference level.
The route erasure operation in TORA floods CLR packets through the network and remove
unacceptable routes, [18].
Figure 3.3
Route Discoveries in TORA – QRY Message
In this figure node A shows and node B shows destination. Node D and node G is one hop
neighbours away from the destination. So they will propagate updates as shown in next figure.
Figure 3.4
Route Discoveries in TORA – Update Message
26
The algorithm [21], contains many flows with most severe in which algorithm extremely
depend upon the numbers of nodes during initial stage, which causes dependence on the
traffic requirement. It is measured as change in rate with regard to total number of nodes
participating in network traffic. In case of gradually positive increase in the network volume,
TORA is not considered as better solution.
27
Chapter 4
Research Methodology
28
4.1
Research Methodology
It illustrates the overall thesis work to be carried out in as in the form of research activity. It
generalizes the understanding about author concept in developing multiple steps or stages for
whole work.
Creswell [22] enlighten two major form of research i.e. quantitative and qualitative.
Quantitative research is carried out by means of inquiry strategies including certain
experiments and simulations which produce some form of statistical data for analysis or
decisions. Qualitative research is based on the researcher knowledge through participatory
and /or constructive perspectives.
4.2
Author approach
Thesis is carried out through multiple stages starting form literature review about the state of
art technology specific to TCP and MANET and then performing the simulation in OPNET
v14.5 tool. So both molds of approaches are necessary to complete the thesis work and have
been administered.
4.3
Research Design
Author believes that the most of the time is spent in identifying and selecting the problem and
conduction literature review. It is also vital to determine that simulation tool also requires
some time during configuration and simulation, as results are often indispensable. Author
divided the overall research into four stages.
4.3.1 Problem Selection / Identification
In problem selection stage author adhere to only MANET and TCP as focus area of research
interest. Problem has been selected in consideration to data performance issues in MANET
due to wireless nature of the network.
4.3.2
Literature Review for TCP and MANET
It is necessary to conduct literature study for understand basic and expertise concept regarding
TCP variants, Ad hoc networks, MANET protocols and related work. OPNET simulator study
is also important to develop the scenario, run simulation and collect result.
29
Internet
State of art
Real
Problem
Technology
Evolution
State of
art study
Literature
Review of
TCP and
MANET
Problem
Identification
and Selection
Author
Interest
Performance
Analysis
Graph
Results
Building
Scenario
Protocol
Selection
Problem
Simulation
Implementation
Simulation
Results
Tool
Selection
Assumption and
Conclusions
Books
TCP Variant
Selection
Figure 4.1
Tool
Configuration
Research Methodology
4.3.3 Building Simulation Environment
This stage is critical to understand as it exhibit deep understanding of how and why
these specific nodes and protocols are chosen. Also it conceptually accelerates the
whole simulation work efficiently and smoothly.
4.3.4 Simulation Results
The last stage is to grasp the relevant results of interest for research study. Simulation
graph results consist of TCP performance for Tahoe, Reno and New Reno relative to
each selectable MANET protocol i.e. AODV, DSR and TORA.
4.4
Simulation Tool
OPNET is selected as it offer easy graphical interface, possibility to develop and run
this simulation environment, validity of the simulation results and the tools is highly
reliable, robust and efficient. Another factor of choosing OPNET is the author’s
familiarity and trust over the tool.
30
Chapter 5
Simulation and Empirical
Study
31
5.1
Simulation Environment
This section is the major portion of the thesis, it is important to setup simulation environment
to observer TCP behavior over MANET. Quantitative analysis is conducted to with the help
of OPNET tool.
5.1.1
Model
Actual problems are simulated through computers by modeling design for actual systems and
theoretical described systems, simulations are executed through computers for the collection
of relative informational results. Learning by doing [23], is a primary concept about systems
in study which requires modeling and operating them.
Model
Design
Execution
Analysis
Model
Execution
Figure 5.1: Simulation Model, [23]
5.1.2
OPNET
Optimized Network Evaluation Tool (OPNET version 14.5) modeler is used in
for simulation network design and execution. OPNET technologies Inc [24],
offers drag and drop communication devices, maximum technologies, multiple
protocols and virtual environment for simulation.
5.1.3
Network Entities and Functions
Simulation network model contain important entities as depicted in the figure
below:
32
Figure 5.2
5.1.3.1
MANET eight node scenarios
Application Definition
In our thesis, this module has been configured to achieve the desired results.
Basically we administer application name as FTP_Application and made
necessary configurations, [Appendix].
5.1.3.2
Profile Configuration
In this configuration I set the profile name with start time offset and starts time,
[Appendix].
5.1.3.3
Mobility Configuration
This module is required to produce the mobility options for the mobile nodes; I
have set default random waypoint and adjusted some parameters, [Appendix].
5.1.3.4
Server
This module controls the traffic for MANET, protocol selection and TCP flavor
has been selected through this module, [Appendix].
5.1.3.5
Mobile Nodes
I have set the mobility options; dynamically allocate the IP addresses to all
mobile nodes including server as well as different routing protocols,
[Appendix].
5.2
Simulation Configuration
A detail configure is described in [Appendix].
5.3
Simulation Flow
Step by step approach is used to keep the simulation flow clear and easy.
Following model for simulation flow is drawn.
33
Apply
Configuration
and Statistics
MANET
Network
design
Running/
Executing
Simulation
Collecting
Results
Figure 5.3: Simulation Flow
MANET network is designed for the simulating the model and configuration statistics
are applied on the whole model separately and the with regard to each individual
scenario.
Simulation is executed multiple times to obtain accurate results describe in each
scenario.
Results from each scenario are gathered as graphs and stored in .jpg files to further use
in the thesis report. Some graphs were plotted one over another and with different
protocol selections and TCP flavors to observe the require statistics.
5.4
Scenarios
The results obtained during the simulation are depicted in this area. In simulation there are
three type of different scenarios based on the number of nodes. Mainly we are concern with
throughput of each variant of TCP and to find out which variants out perform with what type
of routing protocol. To keep clear analysis let us consider each scenario separately, starting
from the three node scenario.
Three nodes Scenario
The table 5.1 shows the detail of routing protocol and TCP variants in three nodes scenario.
Without these three nodes there is a fixed node which working as source. A connection has
been established from each node to transfer a file of the same size over each connection. Then
to find out that which variants of TCP perform well and what was the underlying protocol.
34
Table5.1
Detail of TCP Variants and Routing Protocol for Three Nodes
THROUGHPUT/ DELAY/ CONGESTION WINDOW MEASUREMENT
Stages
A
A
A
B
B
B
C
C
C
TCP Variants
New Reno
New Reno
New Reno
Reno
Reno
Reno
Tahoe
Tahoe
Tahoe
Protocol
AODV
DSR
TORA
AODV
DSR
TORA
AODV
DSR
TORA
Number of nodes
3
3
3
3
3
3
3
3
3
Speed of Nodes m/s
10
10
10
10
10
10
10
10
10
Throughput
From figure 5.4, by looking at the throughput performance between AODV and DSR at
approximately 25 seconds, AODV has better throughput performance over DSR. Comparing
AODV with TORA at the approximation of 95 sec, AODV provides better throughput
performance, in all the three graphs.
Total time consumed by each protocol for send the data through New Reno, Reno and Tahoe
is relatively smaller for AODV followed by DSR and than TORA. It means that TORA has
worst throughput performance in each of the three cases for MANET 3 nodes.The cause
behind the TORA less performance is considered as the reality of TORA working for route
recreation, maintenance and erasure, if dropping of the route occurs, which requires more
time and have bad impact in the data performance.
35
A
B
C
Figure 5.4: Throughput comparison in three nodes scenario
Five Nodes Scenario.
For five nodes scenario the detail of different protocols as shown in Table 5.2. There are five
nodes working as clients to establish connection with a fixed node working as source, and to
transfer a file of the same size over each connection.
Table5.2
Detail of TCP Variants and Routing Protocol for Five Nodes
THROUGHPUT/ DELAY/ CONGESTION WINDOW MEASUREMENT
Stages
A
A
A
B
B
B
C
C
C
TCP Variants
New Reno
New Reno
New Reno
Reno
Reno
Reno
Tahoe
Tahoe
Tahoe
Protocol
AODV
DSR
TORA
AODV
DSR
TORA
AODV
DSR
TORA
Number of nodes
5
5
5
5
5
5
5
5
5
Speed of Nodes m/s
10
10
10
10
10
10
10
10
10
Throughput
Form figure 5.5, comparing all the three graphs with each other I have more or less same
assumption as for node three scenario, but when comparing figure 5.5 with figure 5.4, we
observe that as the number of the nodes are increased approximately 75% (3 to 5), throughput
has been decreased for every TCP variant and each desired protocols.
36
A
B
C
Figure 5.5: Throughput comparison in five nodes scenario
Eight Nodes Scenario
Different protocols evaluated in this scenario are listed in Table 5.3. Like other scenario
depending on the number of clients eight simultaneous connections has been established with
a fixed source to download a file of the same size over each connection.
Table5.3 Detail of TCP Variants and Routing Protocol for Eight Nodes
THROUGHPUT/ DELAY/ CONGESTION WINDOW MEASUREMENT
Stages
A
A
A
B
B
B
C
C
C
TCP Variants
New Reno
New Reno
New Reno
Reno
Reno
Reno
Tahoe
Tahoe
Tahoe
Protocol
AODV
DSR
TORA
AODV
DSR
TORA
AODV
DSR
TORA
Number of nodes
8
8
8
8
8
8
8
8
8
Speed of Nodes m/s
10
10
10
10
10
10
10
10
10
Throughput
From figure 5.6, we conclude that the graph behavior remains same and the assumption made
for figure 5.4 are also acknowledgeable. Further we observe that the AODV have the higher
throughput in the start in all the three figures (5.4, 5.5, and 5.6) as compared to the DSR and
TORA. A small change has been observed in the number of data packets when nodes are
increased to 8 in figure 5.6. The highest number of data packets are reduces from
approximately 750 to 575, which means that if more and more nodes are added in MANET,
37
throughput will reduce. Similarly TORA takes more time when the numbers of nodes are
increased to 8, as compared to 3 nodes and 5 nodes.
A
B
C
Table 5.6 Throughput comparison in eight nodes scenario
Assumptions
It has been concluded that throughput performance of DSR and TORA are minutely affected
with increase in the number of nodes and due to mobility. In general, MANET could have
dynamic number of nodes connectivity in mobility, so it’s important to realize that when the
number of nodes is higher, DSR and TORA would be avoided. AODV has better throughput
performance shown in all three figures as compared to DSR and TORA and is the best
solution for MANET. However simulation results for AODV with respect to New Reno, Reno
and Tahoe depict that throughput is the same in all the cases, so our proposed solution in this
case will be New Reno as it offers multiple packet loss recovery.
Delay
Figure 5.7 hold the simulation results of each and every TCP variant with respect different
routing protocols all together.
With three node scenario, TORA has the highest delay as compared to DSR and AODV
which validate our simulation results. With 5 node and 8 node scenario, TORA and DSR have
approximately same delay, which is not considered as a good perception in wireless networks,
as we encounter numerous losses and delays due to SNR, reflection, diffraction and inter
symbol interference, so the delay measure is considered highly sensitive. AODV on the other
hand have similar delay for 9 results, it maybe because the high resolution graph will show us
small difference in each graph subject to 3, 5 and 8 nodes respectively.
38
1
2
3
4
5
6
7
8
9
Figure 5.7 Delay Comparisons in Different Scenario
39
Congestion Window
Figure 5.8 hold 6 graphs define earlier for all the simulated scenarios; there are much
dissimilarity within each set of variants and routing protocols.
Considering 3 node scenario, congestion window of TORA reaches to the maximum of
160,000 bytes and will remain there which depicts that the exponential increase in the
congestion window leads to the delay and lower performance, as there are not enough space
in the buffer to manage incoming data and the window remains constant.
1
2
3
4
5
6
Figure 2.8 Congestion Window dynamics
Secondly, DSR have different results in all the six cases, variations for congestion window are
found approximately between 65,000 bytes to 80,000 except for case 3, where it had reached
around 11,000 bytes. So it’s likely to know that DSR congestion window is quite uncertain
40
and its use in the MANET network will have severe results. Congestion window will
ultimately influence throughput and delay parameters important for the performance
evaluation.
Third protocol, AODV have similar results during individual 3 nodes and 5 nodes (New
Reno, Reno and Tahoe). It explicitly defines that AODV has the formal behavior of TCP as
within the normal networks. The congestion window is dynamically increasing and
decreasing after receiving reply for the receiver. It is however not much clear whether it
doubles the size of the congestion window after each successful data packet transmission or
not, but the behavior of the graph exactly resembles with the actual TCP congestion window
graph with increase and decrease. We suggest AODV as the best possible solution and
recommendation for MANET.
41
Chapter 6
Discussions/ Results
42
6.1
Conclusion
The overall research has been conducted in the well organized way as defined in chapter 4.
Research questions 1 and 2 defined in chapter 1 are answered through state of art study and
literature review. Research question 3 and four are based on the simulation results and
explanations.
We adhere to the simulation results as an evident that TCP variants have minor affect on the
overall results except in few cases defined earlier, but the major dependence lies on MANET
routing protocols. Simulation observation based on AODV, DSR and TORA clearly describe
us about the performance evaluation through measuring throughput, delay and congestion
window that the best routing protocol for MANET is AODV.
6.2
Future Work
As I have selected these numerous MANET routing protocols of interest by simulation in an
OPNET tool, another possibility of doing the same work can be done through another tool
like NS-2. Also, selection of other routing protocols can be use for the performance
evaluation or other parameters of performance could be considered for simulation.
43
Appendix: Simulation Steps
Open the OPNET simulator.
Click on File and select New then click ok
Enter name to project and scenario then click ok.
Select Create empty scenario and click Next
44
Select Campus Network then Click Next
X span: 1000
Y span: 1000
Units : Meters then Click Next
Select the Technology (MANET) then Click Next
45
Now Click Finish
From object Palette Tree select one by one
Application Config, Profile Config, Mobility Config, Wlan_Server (Fixd Node) and
Wlan_wkstn (Mobile node) with the requirements of the user.
46
Drag all these object one by one in the working area.
Right Click on Application Config, profile Config, mobility Config and Server to set
their name.
47
Click Ok
1 Application Configuration
Right Click on Application Config then Edit Attributes and configure the Application.
48
49
At last Click Ok
2 Profile Configurations
Right Click on Profile Config then Edit Attributes and configure the Profile.
50
Select all Nodes and Server Click on Protocol menu  IP  Addressing  AutoAssign IPv4 Address.
Select all Nodes and Server  Right Click on Server  Edit Attributes  Select ADHOC Routing Protocol .e.g. AODV.
In TCP Select the Parameters e.g. Tahoe.
Tick Apply to Selected Objects  Ok
51
3 Server Configurations
Right Click on Server  Edit Attributes  Application  set Application Supported
Services and Click Ok.
52
4 Mobile Node Configurations
Select all Nodes Right Click on one  Edit Attributes. Click on Applications 
Application: Destination Preferences and set some parameters.
Click on Application: Supported Profiles and set some values  Tick Apply to
Selected Objects. Click Ok.
53
54
5 Click Topology menu  Select Random Mobility  set Mobility Profile  Click
Continue  Click Ok.
6 Mobility Configurations
Right Click on Mobility Config  Edit Attributes  Click Default Random Waypoint
 Click Random Waypoint Parameters. Click OK.
55
7 Save the Project.
8 Click DES Menus
Click DES or Right Click on work area and select Individual Statistics.
56
Expand Node Statistics
a) Expand TCP Connection and Tick the Congestion Window Size (byte).
Right Click on it then choose Change Collection Mode then Tick Advance and
Select All Values then Click Ok.
b) Expand Wireless Lan Tick Delay (sec) and Throughput (bit/sec) then Right Click
on both one by one and select Change Collection Mode then Tick Advance and
Select All Values then Click Ok.
9 Run the Project
Click on DES menu Select Configure/Run Discrete Event Simulation ….ctrl+R.
Here you can set Time in Weeks, Days, Hours, Minuets, and Seconds. Then Click
Run.
57
When Simulation is Completed then Click Close.
58
10 Show the Result
Click on DES menu or Right Click on Work area then Select Results then View result
59
References:
[1] http://www.ietf.org/rfc/rfc1180.txt
[2] http://www.ietf.org/rfc/rfc2501.txt
[3] K. Leung and Vivtor O.K. Li, “Transmission Control Protocol (TCP) in wireless
Networks: issues, approaches and challenges,” IEEE Communications Survey, Vol. 8 No.
4, pp. 64-79, 4th October 2006
[4] IEEE Standard 802.11, Wireless LAN Medium Access Control (MAC) and Physical
Layer (PHY) Specifications, June 2007
[5] A. A. Hanbali, E. Altman, and P. Nain, “A Survey of TCP over Ad Hoc Networks,” IEEE
Communication. Surveys and Tutorials, Vol. 7, No. 3, 3rd Quarter 2005,
[6] J. Postal,”Transmission Control Protocol”, Request for comments, RFC 793, September
1981.
[7] V. Jacobson, “Congestion avoidance and control,” in: Proceedings of ACM SIGCOMM,
pp. 314-29, 1988.
[8] Allman, V. Paxson, and W. Stevens, “TCP Congestion Control,” Request for Comments,
RFC 2581, Network Working Group, Internet Engineering Task Force, April. 1999.
[9] Jany Hoe: S. Floyd and T. Henderson, “The NewReno Modification to TCP's Fast
Recovery Algorithm,” RFC 2582, Network Working Group, Internet Engineering Task
Force, April. 1999.
[10] D. Kliazovich, F. Granelli, “C3TCP: Cross-layer congestion control in ad hoc wireless
networks”, ELSEVIER Ad HocNetworks 4, pp. 687-708, 2006.
[11] J. Liu and S. Singh, “ATCP: TCP for Mobile ad hoc Networks”, IEEE Journal in
Selected Areas in Communication (JSAC), Vol 19, No.7, pp. 1300-15, July 2001,.
[12]
R. Cheng, H. Lin, “A Cross-layer Design for TCP End-to-End Performance
Improvement in multi-hop wireless networks”, ELSEVIER, Computer communication
Vol. 31, issue 14, pp. 3145-3152, Sep. 2008
[13] Routing and Multicasting Strategies in Wireless Mobile Ad hoc Networks by Sung –Ju
Lee University of California, Los Angeles 2000.
[14] A. Zahary, A. Ayesh, “Analytical study to detect threshold number of efficient routes in
multipath AODV extensions”, proceedings of International Conference of Computer
Engineering & Systems, ICCES, 2007, pp. 95 – 100
[15] R. Misra and C.R. Mandal, “Performance comparison of AODV/DSR on-demand
routing protocols for ad hoc networks in constrained situation” ICPWC International
Conference, IEEE, 2005, pp. 86 – 89
[16] Lijuan Cao, K. Sharif, Yu Wang, T. Dahlberg, “Adaptive Multiple Metrics Routing
Protocols for Heterogeneous Multi-Hop Wireless Networks”,
[17] Changling Liu, Jörg Kaiser, A Survey of Mobile Ad Hoc network Routing Protocols*
kaiser@ivs.cs.uni-magdeburg.de
[18] Jin Mook Kim, In Sung Han, Jin Baek Kwon, Hwang Bin Ryou, “A Novel Approach to
Search a Node in MANET”, Information Science and Security, ICISS, 2008, pp. 44 – 48
60
[19] R. Thorulp, “Mobile Ad Hoc Networks and Routing Protocols”, Implementing and
Evaluating the DYMO Routing Protocol, Master’s Thesis at the University of AARHUS,
pp. 7- 20, 2007
[20] Jin Mook Kim, In Sung Han, Jin Baek Kwon, Hwang Bin Ryou, “A Novel Approach to
Search a Node in MANET”, Information Science and Security, ICISS, 2008, pp. 44 – 48
[21] S. R. Chaudhry, A. Al-Khwildi, Y. Casey, H. Aldelou, “A Performance Comparison of
Multi On- Demand Routing in Wireless Ad Hoc Networks”, Wireless Network and
Communications Group (WNCG), School of Engineering & Design, Brunel University,
West London
[22] J. W. Creswell, Research Design: Qualitative, Quantitative and Mixed Methods
Approach, 2nd Ed, Sage Publications Inc, California, July 2002.
[23] http://www.cis.ufl.edu/~fishwick/introsim/node1.html
[24] http://www.opnet.com/
61