|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
Performance Evaluation of Mobile Ad Hoc Ne tworks in Realistic Mobility and Fading Environments by Preetha Prabhakaran A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Electrical Engineering Department of Electrical Engineering College of Engineering University of South Florida Major Professor: Ravi Sankar, Ph.D. Miguel Labrador, Ph.D. Arthur David Snider, Ph.D. Date of Approval: March 24, 2005 Keywords: ns-2, rayleigh fading, mobility mode ls, packet delivery ratio, control overhead Copyright 2005 Preetha Prabhakaran
Dedication To my Parents and Sister for their everlasting love.
Acknowledgments I would like to convey my sincere thanks to Dr. Ravi Sankar for being very supportive and encouraging all through my thesis work. I would also like to thank Dr. Miguel Labrador and Dr. Arthur David Snider for their guidance in my work. I would like to express my love and sincere appreciation to my parents, Prabhakaran and Shyamala; my sister, Prathibha; and Rakesh, who mean the world to me. Their strong belief in me has really helped me be the person I am today. I would like to convey my affection and gratitude to Kumar for being there for me at any point of time and fo r his wonderful friendship. I would also like to thank Sivakumar Bakthavachal u for all the support and guidance he offered to me when I was learning ns-2 simulator. Last but not the least I would like to thank all my friends and my room mates at USF for their moral support and encouragement.
i Table of Contents List of Tables iv List of Figures v Abstract x Chapter One: Introduction 1 1.1 Wireless LANs 1 1.1.1 Working of WLANs 1 1.1.2 Wireless LAN Technology Options 2 1.1.3 Classification of WLANs 4 1.2 Mobile Ad hoc Networks 5 1.3 Research Challenges of MANETs 7 1.3.1 Throughput 7 1.3.2 Multi-path Fading 9 1.3.3 Energy Utilization 9 1.3.4 Mobility 10 1.3.5 Scalability 11 1.4 Motivation 12 1.5 Research Objectives 13 1.6 Thesis Organization 13 1.7 Summary 14 Chapter Two: Literature Review and Background 15 2.1 Literature Review 15 2.2 Fading Channel 16 2.3 IEEE 802.11 Medium Access Control (MAC) 18
ii 2.4 Routing Protocols 20 2.4.1 Dynamic Source Routing (DSR) Protocol 20 2.4.2 Ad hoc On-Demand Distan ce Vector (AODV) Ro uting Protocol 21 2.4.3 Destination-Sequenced Dist ance Vector (DSDV) Routing Protocol 22 2.5 Mobility Models 22 2.5.1 Random Waypoint Model (RW) 22 2.5.2 Gauss Markov Model (GM) 23 2.5.3 Manhattan Grid Model (MG) 24 2.5.4 Reference Point Group Mobility (RPGM) Model 25 2.5.5 Pursue Model 26 2.5.6 Column Model 26 2.6 Summary 27 Chapter Three: Factors Infl uencing MANET Performance in Realistic Environments 28 3.1 Network Simulations in Realistic Environments 28 3.2 Temporal Dependency of Velocity 29 3.3 Spatial Dependency of Velocity 30 3.4 Geographic Restriction of Movements 31 3.5 Effect of Multipath Fading Channel 32 3.6 Scalability of a Network 33 3.7 Summary 33 Chapter Four: Network Simulation Environment 34 4.1 ns-2 (Network Simulator) 34 4.1.1 Origins 34 4.1.2 Functional Description 34 4.1.3 Modifications to ns-2 36 4.2 Mobility Generators 37 4.2.1 The setdest Mobility Generator 37 4.2.2 BonnMotion Mobility Generator 38
iii 4.2.3 Scengen Mobility Generator 40 4.2.4 Mobility Generator Toilers Code 40 4.3 Traffic Generation 43 4.4 Network Scenarios 43 4.5 Energy Consumption Model 45 4.6 Performance Evaluation 46 4.7 Summary 46 Chapter Five: Experimental results 47 5.1 Scalability in Mobile Ad hoc Networks 47 5.1.1 Scalability Analysis of MANETs using Entity Mobility Models 47 5.1.2 Scalability Analysis of MANETs using Group Mobility Models 52 5.2 Energy Utilization in Mobile Ad hoc Networks 71 5.2.1 Energy-Goodput Analysis of MANETs using Entity Mobility Models 71 5.2.2 Energy-Goodput Analysis using Group Mobility Models 73 5.3 Summary 81 Chapter Six: Conclusions and Future Work 82 6.1 Conclusions 82 6.2 Future Work 84 References 85
iv List of Tables Table 3.1 Mobility Models and their Movement Characteristics 32 Table 4.1 Multi-path Fading Model Parameters 37 Table 4.2 Parameters used to generate Manhattan Gr id Movement Pattern 38 Table 4.3 Parameters used to generate Gauss Mar kov Movement Pattern 39 Table 4.4 Parameters used to generate Pursue Moveme nt Pattern 41 Table 4.5 Parameters used to generate Gauss Mar kov Movement Pattern 42 Table 4.6 Different RPGM Scenarios Simulated 43 Table 4.7 Different Group Scenarios Simulated 44 Table 4.8 Power Consumption Values 46 Table 5.1 Maximum Control Overhead Produced for Single and Multiple Group Scenarios when Implementing RPGM Mobility Model 60 Table 5.2 Maximum Control Overhead Produced for Single and Multiple Group Scenarios when Implementing Column Mobility Model 66
v List of Figures Figure 1.1 Typical WLAN Configuration 2 Figure 1.2 Direct-Sequence Spread Spectrum 3 Figure 1.3 Frequency Hoppi ng Spread Spectrum 4 Figure 1.4 Schematic of a WLAN Infrastructure Network 5 Figure 1.5 Schematic of a WLAN Ad hoc Network Single hop and Multi-hop Configurations 5 Figure 1.6 Hidden Terminal Scenario 8 Figure 1.7 Exposed Terminal Scenario 8 Figure 2.1 Illustration of Fading Mechanisms 17 Figure 2.2 Multi-path or Rayleigh Fading 17 Figure 2.3 Schematic of the IEEE 802.11 Netw ork Architecture 19 Figure 2.4 Traveling Pattern of a M obile Node using Random Waypoint Mobility Model 23 Figure 2.5 Traveling Pattern of a Mobile Node using Gauss Markov Mobility Model 24 Figure 2.6 Traveling Pattern of a Mobile Node using Manhattan Grid Mobility Model 24 Figure 2.7 Traveling Patterns of Th ree Mobile Nodes using Reference Point Group Mobility Model 25 Figure 2.8 Movement of Mobile Nodes using Pu rsue Mobility Model 26 Figure 2.9 Movement of Mobile Nodes using Column Mobility Model 27 Figure 3.1 Network Simulation in an Unreal istic Environment 29
vi Figure 3.2 Network Simulation in a Realistic Envi ronment 30 Figure 3.3 Pathway Map of Manhatta n Grid Mobility Model 31 Figure 3.4 Time-Sequenced Rayleigh Fading Envelope 32 Figure 4.1 Schematic of a Mobilenode under the CMU Monarch's Wireless Extensions to Ns-2 36 Figure 4.2 Network Schematic Showing Initial N ode Positions 44 Figure 4.3 Illustration of Group Scenario 3 45 Figure 5.1 PDR Analysis for Various Routing Algorithms using Random Waypoint Mobility Model 48 Figure 5.2 PDR Analysis for Various Routing Algorithms using Manhattan Grid Mobility Model 49 Figure 5.3 PDR Analysis for Various Routing Algorithms using Gauss Markov Mobility Model 49 Figure 5.4 Control Overhead Analys is for Various Routing Algorithms using Random Waypoint Mobility Model 50 Figure 5.5 Control Overhead Analys is for Various Routing Algorithms using Manhattan Grid Mobility Model 51 Figure 5.6 Control Overhead Analys is for Various Routing Algorithms using Gauss Markov Mobility Model 51 Figure 5.7 PDR Analysis for DSR using RPGM Mobility Model for Various Speeds 52 Figure 5.8 PDR Analysis for DSR using RPGM Mobility Model for Various Group Scenarios 53 Figure 5.9 PDR Analysis for AODV using RPGM Mobility Model for Various Speeds 53 Figure 5.10 PDR Analysis for AODV using RPGM Mobility Model for Various Group Scenarios 54 Figure 5.11 PDR Analysis for DSDV using RPGM Mobility Model for Various Speeds 54
vii Figure 5.12 PDR Analysis for DSDV using RPGM Mobility Model for Various Group Scenarios 55 Figure 5.13 Control Overhead Analysis for DSR using RPGM Mobility Model for Various Speeds 56 Figure 5.14 Control Overhead An alysis for AODV using RPGM Mobility Model for Various Speeds 56 Figure 5.15 Control Overhead Analysis for DSDV using RPGM Mobility Model for Various Speeds 57 Figure 5.16 Control Overhead Analysis for DSR using RPGM Mobility Model for Various Group Scenarios 58 Figure 5.17 Control Overhead An alysis for AODV using RPGM Mobility Model for Various Group Scenarios 59 Figure 5.18 Control Overhead Analysis for DSDV using RPGM Mobility Model for Various Group Scenarios 59 Figure 5.19 PDR Analysis for DSR us ing Column Mobility Model for Various Speeds 60 Figure 5.20 PDR Analysis for DSR using Column Mobility Model for Various Group Scenarios 61 Figure 5.21 PDR Analysis for AOD V using Column Mobility Model for Various Speeds 61 Figure 5.22 PDR Analysis for AOD V using Column Mobility Model for Various Group Scenarios 62 Figure 5.23 PDR Analysis for DSDV using Column Mobility Model for Various Speeds 62 Figure 5.24 PDR Analysis for DSDV using Column Mobility Model for Various Group Scenarios 63 Figure 5.25 Control Overhead Analysis for DSR using Column Mobility Model for Various Speeds 64 Figure 5.26 Control Overhead Analysis for DSR using Column Mobility Model for Various Group Scenarios 65
viii Figure 5.27 Control Overhead An alysis for AODV using Column Mobility Model for Various Speeds 66 Figure 5.28 Control Overhead An alysis for AODV using Column Mobility Model for Various Group Scenarios 67 Figure 5.29 Control Overhead Analysis for DSDV using Column Mobility Model for Various Speeds 67 Figure 5.30 Control Overhead Analysis for DSDV using Column Mobility Model for Various Group Scenarios 68 Figure 5.31 PDR Analysis for Various Routing Algorithms using Pursue Mobility Model 68 Figure 5.32 Control Overhead Analysis for Various Routing Algorithms using Pursue Mobility Model 69 Figure 5.33 Energy-Goodput Analysis of Various Routing Algorithms using Random Waypoint Mobility Model 71 Figure 5.34 Energy-Goodput Analysis of Various Routing Algorithms using Manhattan Grid Mobility Model 72 Figure 5.35 Energy-Goodput Analysis of Various Routing Algorithms using Gauss Markov Mobility Model 73 Figure 5.36 Energy-G oodput Analysis of DSR using RPGM Mobility Model for Various Speeds 74 Figure 5.37 Energy-Goodput Analysis of DSR using RPGM Mobility Model for Various Group Scenarios 74 Figure 5.38 Energy-Goodput Analysis of AODV using RPGM Mobility Model for Various Speeds 75 Figure 5.39 Energy-Goodput Analysis of AODV using RPGM Mobility Model for Various Group Scenarios 76 Figure 5.40 Energy-Goodput Analysis of DSDV using RPGM Mobility Model for Various Speeds 76 Figure 5.41 Energy-Goodput Analysis of DSDV using RPGM Mobility Model for Various Group Scenarios 77
ix Figure 5.42 Energy-Goodput Analysis of DSR using Column Mobility Model for Various Speeds 77 Figure 5.43 Energy-Goodput Analysis of DSR using Column Mobility Model for Various Group Scenarios 78 Figure 5.44 Energy-Goodput Analysis of AODV using Column Mobility Model for Various Speeds 78 Figure 5.45 Energy-Goodput Analysis of AODV using Column Mobility Model for Various Group Scenarios 79 Figure 5.46 Energy-Goodput Analysis of DSDV using Column Mobility Model for Various Speeds 79 Figure 5.47 Energy-Goodput Analysis of DSDV using Column Mobility Model for Various Group Scenarios 80 Figure 5.48 Energy-Goodput Analysis of Various Routing Algorithms using Pursue Mobility Model 80
Performance Evaluation of Mobile Ad Hoc Networks in Realistic Mobility and Fading Environments Preetha Prabhakaran ABSTRACT Mobile Ad hoc Networks (MANETs) are wireless networks, which consist of a collection of mobile nodes with no fixed infrastructure, where each node acts as a router that participates in forwarding data packets. They are a new paradigm of wireless communications for mobile hosts that are resource-constrained with only limited energy, computing power and memory. Previous studies on MANETs concentrated more on energy conservation in an idealistic environment without taking into consideration, the effects of realistic mobility, interference and fading. The definition of realistic mobility models is one of the most critical and, at the same time, difficult aspects of the simulations of networks designed for real mobile ad hoc environments. The reason for this is that most scenarios for which ad hoc networks are used have features such as dynamicity and extreme uncertainties. Thus use of real life measurements is currently almost impossible and most certainly expensive. Hence the commonly used alternative is to simulate the movement patterns and hence the reproduction of movement traces quite similar to human mobility behavior is extremely important. The synthetic models used for movement pattern generation should reflect the movement of the real mobile devices, which are usually carried by humans, so the movement of such devices is necessarily based on human decisions. Regularity is an important characteristic of human movement patterns. All simulated movement models are suspect x
because there is no means of accessing to what extent they map reality. However it is not difficult to see that random mobility models such as Random Walk, Random Waypoint (default model used in almost all network simulations), etc., generate movements that are most non-humanlike. Hence we need to focus on more realistic mobility models such as Gauss Markov, Manhattan Grid, Reference Point Group Mobility Model (RPGM), Column, Pursue and other Hybrid mobility models. These models capture certain mobility characteristics that emulate the realistic MANETs movement, such as temporal dependency, spatial dependency and geographic restriction. Also a Rayleigh/Ricean fading channel is introduced to obtain a realistic fading environment. The energy consumed by the data, MAC, ARP and RTR packets using IEEE 802.11 MAC protocol with the various mobility models in fading and non-fading channel conditions are obtained using ns-2 simulations and AWK programs. The realistic movement patterns are generated using three different mobility generators BonnMotion Mobility Generator, Toilers Code and Scengen Mobility Generator. This thesis work performs an in-depth study on the effects of realistic mobility and fading on energy consumption, packet delivery ratio and control overhead of MANETs. xi
1 Chapter One Introduction 1.1 Wireless LANs Wireless networking is an exciting technology that enables two or more computers to communicate using standard network protocols, but w ithout network cabling. WLAN (Wireless Local-Area Network) is a category of local-area network that uses highfrequency radio waves rather th an wires to communicate between nodes such as computers, Internet devices or other applia nces. It is a flexible data communication system implemented as an extension to or as an alternative for, a wired LAN within a building or campus. Wi-Fi networks use radio t echnologies called IEEE 802.11b or 802.11a to provide secure, reliable, fast wi reless connectivity. A Wi-Fi network can be used to connect computers to each other, to the Internet, and to wired networks (which use IEEE 802.3 or Ethernet ) . Wi-Fi networks operate in the unlicensed 2.4 and 5 GHz radio bands, with an 11 M bps (IEEE 802.11b) or 54 Mbps (IEEE 802.11a) data rate or with products that contain both bands (dual band), so they can provide real-world performance similar to the basic 10BaseT wired Ethernet ne tworks used in many work places, IEEE 802.11 being the IEEE standard for WLANs. WLANs are becoming more important due to increased interest in connection of mobile and portable computers mutually, or to the wired LANs. 1.1.1 Working of WLANs Wireless LANs use electromagnetic waves to communicate information from one point to another without relying on any physical connection. WLANs combine data connectivity with user m obility and enables movable LANs through simplified configurations. Radio waves are often referred to as radio carriers because they
2 simply perform the function of delivering in formation to a remote receiver. Data is modulated onto a radio carrier and then tran smitted. At the radio receiver the data is extracted from the modulated signal by demodulation . Figure 1.1. Typical WLAN Configuration In a WLAN configuration, as shown in Figure 1.1, an access point (AP), which is a transceiver device, connects to a wired netw ork using a standard Ethernet cable. A single AP can function within a certain range and can support a sm all group of users. The AP receives, buffers, and transmits data between the WLAN and the wired network infrastructure. End users acce ss the WLAN through wireless LAN adapters, which are implemented as PC cards in not ebook computers or use fully integrated devices within handheld computers . 1.1.2 Wireless LAN Technology Options There are various WLAN technology opt ions. Each has its own advantages and disadvantages . They are: Direct-Sequence Spread Spectrum Technology: Figure 1.2 represents a Direct Sequence Spread Spectrum (DSSS) which gene rates a redundant bit pattern for each
3 bit to be transmitted. This bit pattern a chip or chipping code. The longer the chip, the greater the probability that the original data can be recovered. Even if one or more bits in a chip are damaged during transmi ssion, statistical techni ques embedded in the radio can recover the origin al data without the need for retransmission. To an unintended receiver, DSSS appears as low-pow er wideband noise and is rejected by most narrowband receivers. Figure 1.2. Direct-Sequence Spread Spectrum Frequency Hopping Spread Spectrum Technology: Figure 1.3 represents a Frequency Hopping Spread Spectrum (FHSS) which uses a narrowband carrier that changes frequency in a pattern known to both transmitter and receiver. To an unintended receiver, FHSS appears to be short duration impulse noise. Narrowband Technology: A narrowband radio system transmits and receives user information on a specific radio frequency. Narrowband radio keeps the radio signal
4 frequency as narrow as possible just to pass the information. Undesirable crosstalk between communications channels is avoi ded by carefully coordinating different users on different channel frequencies. Figure 1.3. Frequency H opping Spread Spectrum In a radio system, privacy and non-interference are accomplished by the use of separate radio frequencies. The radio receiver filters out all the radio signals except the ones on its designated frequency. 1.1.3 Classification of WLANs Infrastructure Network: In an infrastructure network, as shown in Figure 1.4, the wireless devices communicate with a central node that in turn can communicate with wired nodes on that LAN. They are compri sed of WLANs connect ed to wired LANs and contain access points to channel network traffic. Ad hoc Network: Ad hoc network is comprised of wireless devices that communicate with each other in a peer-to-peer mode. Ad -hoc mode is useful for establishing a network where wireless infr astructure does not exist. Figure 1.5 depicts a single hop
5 and multi-hop ad hoc network configurations In single hop ad hoc mode, there is no routing operation and hence only one-to-o ne communication, while in a multi-hop adhoc mode the network nodes communicate via other nodes. Figure 1.4. Schematic of a WLAN Infrastructure Network Figure 1.5. Schematic of a WLAN Ad hoc Network Single hop and Multi-hop Configurations
6 1.2 Mobile Ad hoc Networks (MANET) Mobile and wireless technology is growing at a rapid rate. Ad hoc networks are a consequence of the ceaseless research effort s in mobile and wireless networks. They are a new paradigm of wireless communica tions for mobile hosts. Each MANET is a set of wireless mobile hosts forming a temporary, dynamic autonomous network without the aid of any established infrastruc ture or centralized ad ministration, such as base stations or mobile switching centers . Wireless technologies such as General Pack et Radio Service, Wi-Fi, Home-RF, and Bluetooth make it possible to access the Web from mobile phones, print documents from PDAs, and synchronize data among va rious office devices. However, such applications rely at some point on mobility support routers or base stations, and it is often necessary to establish communicatio n when the wired infrastructure is inaccessible, overloaded, damaged, or destroyed . However, MANETs do not rely on any fixed infrastructure but communicate in a selforganized way. Nodes communicate with ea ch other without the intervention of centralized access points or base stations All nodes share the responsibility of network formation and management. Also they all behave as routers and take part in the discovery and maintenance of routes to other nodes in the network. In areas in which there is little or no communication infrastructure, or the existing infrastructure is expensive or inconvenient to use, wireless mobile users may still be able to communicate through the formation of an ad hoc networks. Their main advantages are the lower costs, inherent scalability, portability, mobility, ease of installation and their suitability to free un licensed spectrum . Ad hoc networks have received growing research attention during the last decade. This is partly due to significant developments in local area wire less technologies that are now starting to enable low cost wireless network build-o ut for local area communications. Such networks are emerging first of all in co rporate environments especially through WLAN technologies, and similarly WLAN is al so taking more and more footprint in
7 residential solutions. They are applied most commonly in situations such as military tactical operations, emergency cases, ta rget tracking, law enforcement, rescue missions during disaster, virtual classrooms, and conferences . The strength of ad hoc networks resides in the diversity of computer networking and the growth of wireless over IP that patche s the Internet together. Ad hoc networks are seen as the potential market for embedded network devices in multiple environments such as vehicles, mobile telephones and pers onal appliances. They are considered the infrastructure-less that will allow the user s to create their Personal Area Networks (PANs) . Wireless Personal Area Networks (WPANs) ar e short-range wireless networks that permit communication between wireless devi ces at a distance of around 10 meters. Bluetooth is a WPAN standard used for s hort distance transmissi on of digital voice and data that supports both, point-to-point and multipoint app lications between mobile phones, computers, personal digital a ssistants (PDAs) etc,. It transmits in the unlicensed 2.4 GHz band and uses the fre quency hopping spread spectrum technique. IEEE 802.15 is the working group of IEEE th at develops standard protocols and interfaces for WPANs. 1.3 Research Challenges of MANETs 1.3.1 Throughput One of the fundamental challenges in MANE Ts research is how to increase the overall network throughput while maintain ing low energy consumption for packet processing and communications. The low th roughput is attributed to the harsh characteristics of the radio channel combin ed with the contention-based nature of medium access control (MAC) protocols comm only used in MANETs. The notorious near-far problem undermines the throughput performance in MANETs . Further, concurrent wireless transmi ssions in an ad-hoc network limit its thr oughput capacity, because they create mutu al interference .
8 When two mutually out of range hosts compete over a common host, undetectable receiver side collisions result. When two mutually out of range hosts compete over a common host, undetectable receiver side collis ions result. In other words, due to the limited transmission range of mobile stati ons, multiple transmitters within range of the same receiver may not know one anothers transmissions, and hence are in effect hidden from one another. When these tran smitters transmit to the same receiver at around the same time, they do not realize that their transmissions collide at the receiver. This so-called hidden term inal problem degrades the throughput significantly. The near-far SNR problem ha s a significant effect on the performance of an ad-hoc network. It causes collisi ons which results in loss of efficiency (reduction of throughput). One solution might be to use RTS/CTS signaling, but it may not help much in a multi-hop ad hoc network due to the difference between the transmission range and sensing/interference range and also due to the fact that it increases control signal overhead . Figure 1.6. Hidden Terminal Scenario Figure 1.6 represents the hidden te rminal problem  where: A talks to B C senses the channel C does not hear As transmission (out of range) C talks to B Signal from A and B collide Causes wastage of resources, mainly throughput Another problem related to segment overlapping in ad hoc networks is the exposedterminal problem . In this case the problem arises when the sensing mechanism
9 prevents parallel transmission, from two or mo re terminals, toward receivers that would not observe collision as the receivers are lo cated far apart . In exposed terminal scenario the free channel is not used, resulting in loss of efficiency. Hence hidden terminal problem and exposed node problem are conflicting . Figure 1.7. Exposed Terminal Scenario Figure 1.7 illustrates the exposed terminal problem  where: B talks to A C wants to talk to D C senses channel and finds it to be busy C stays quite (when it could have ideally transmitted) Lower effective throughput due to underutilization of channel 1.3.2 Multi-path Fading This is caused by multipath propagation of radio frequency (RF) signals between a transmitter and a receiver. Multipath propagation can lead to fluctuations in the amplitude, phase, and angle of the signal rece ived at a receiver. If there is a strong LOS (Line Of Sight) between the transmitter and the receiver, diffraction and scattering are not the domina nt factors in the propaga tion of the radio waves. However, in the absence of a LOS between the transmitter and the receiver, diffraction and scattering become the domina nt factors in the propagation. Typically, the received signal is a sum of the co mponents arising from the above three phenomena. The strength of the received signal fluctuates rapidly with respect to time and the displacement of the tran smitter and the receiver .
10 A fundamental characteristic of mobile wireless networks is the time variation of the channel strength of the underlying communica tion links. Such time variation occurs at multiple time scales and can be due to multipath fading, path loss via distance attenuation, shadowing by obstacl es, and interference from other users. The impact of such time variation on the design of wire less networks parameters throughout the layers, ranging from coding and power cont rol at the physical layer to cellular handoff and coverage planning at the networking layer . 1.3.3 Energy Utilization MANETs face power problems because of a lot of reasons  such as, Battery power is limited Recharging or replacing batte ries may be difficult. Large relay traffic in multi-hop routing might cause faster depletion of the node power source. Increased battery size increases the size and weight of the node, while decreased battery size re sults in less capacity. Consumption of battery charge increases with an increase in the transmission power. Power control in MANETs has recently rece ived a lot of attention for two main reasons. First, power control has been shown to increase spatial channel reuse, hence increasing the overall (aggregate) channel utilization. This issu e is particularly critical given the ever-increasing demand for channe l bandwidth in wireless environments. Second, power control improves the overa ll energy consumption in a MANET, consequently prolonging the lifetime of the network . Portable devices are often powered by batteries with limited weight a nd lifetime, and energy saving is a crucial factor that impacts the survivability of such devices. Energy efficiency is one of the most important aspects in mobile networks. Power is arguably the scarcest resource for mobile de vices, and power savi ng has always been a major design issue for the developers of mobile devices, wire less communication
11 protocols and mobile computing systems. It is not practically possi ble to recharge or change batteries every other hour, or to carry a heavy battery pack, power puts many limitations on operations of a mobile device . Com puting ability is sacrificed because high performance processor needs more power. Also transmission range and bandwidth are restricted due to the fact th at long range high bandwidth transceivers consume much energy. 1.3.4 Mobility An ad hoc network consists of nodes that communicate w ith each other without the help of pre-existing infrastructure. The links between the nodes may change and the network adapts rapidly to the new situati on. The freedom of movement makes wireless communication very attractive. But at the same time mobility brings challenges owing to bandwidth and power constraints, limited or no infrastructure and mobility of users. When a link between two nodes that is in use disconnects, the routi ng protocol needs to adapt to the new situation. This creates a cost both in the amount of control traffic and in the message delay i.e., frequent route changes due to mobili ty of the nodes would increase the signaling overhead and end-to-e nd delay which is required to establish a route . When signaling overhead increas es the energy consumed by the network will in turn increase which leads to a redu ced network lifetime. Also because of the quick topology changes due to mobility of the nodes, ordinary routing protocol fails to give good performance. Since nodes are mob ile most of the time a lot of undesirable effects such as disconnection, bit errors, reduction in thro ughput, etc. take place In order to evaluate the impact of mobility wh ile simulating a MANET routing protocol, it is crucial that the underlying mobility model accurately emulates real-world node mobility or at least the essential characteristics. 1.3.5 Scalability Over the last decade, many mobile ad hoc routing protocols have continually been designed and refined. However, most of the designs have been for small to mid size networks, especially those w ith low node density. As a re sult, most mobile ad hoc
12 routing protocols suffer large performance degradations when used in large-scale networks. Performing route discovery in a large or high-density network using reactive protocols, for instance, can be expensive due to network-wide br oadcast floods. Using a proactive protocol in a highly mobile network, on the other hand, also causes significant performance degradations due to large amount of resource spent on updating the routing tables. While many proposed optim izations have been done to mitigate the shortcomings of ad hoc routing protocols in large-scale networks, most of the proposals were designed to address problems related to specific routing protocols in specific environments . Wireless communication systems for military a nd commercial infrastructures have been significantly scaled up in their sizes as well as complexities. Such systems are analytically intractable and simulation is a common alternative to explore the behavior of large-scale, complex wireless netw ork systems. However, existing popular simulation tools such as OPNET  or ns-2 , which have contributed to the wireless communication community in the de sign and evaluation of new protocols, are not capable of simulating large-scale networ k models as the execution times of those simulation tools can be unr easonably long. Moreover, th e memory requirement to simulate such systems physically limits the maximum number of network nodes with the existing simulation tools . 1.4 Motivation A majority of the previous studies on MANETs concentrated on energy utilization, throughput, scalability and packet drop rate in an idealistic environmen t without giving much importance to the effects of realisti c mobility and fading. A lot of interesting details about routing energy overheads of various ad ho c routing protocols were obtained from . The mobility models were classified into entity and group mobility models . Reference  studied the effect of RPGM on various performance metrics such as throughput and contro l overhead. In  the effect of random waypoint model, RPGM model and Manhattan grid model on the overall energy consumption
13 was compared. But energy-goodput, network li fetime, packet drop rate, throughput performance, scalability, etc were not analy zed in detail in the pr esence of fading with realistic mobility models. There have been many schemes to extend the use of mobile ad hoc routing protocols to e nvironments much larger than the traditionally small and low-density ad hoc networks. Such larg e-scale mobile ad hoc networks are characterized by high node densit y and high mobility . The commonly used free space model is comp utationally efficient but ignores many losses that are common in wireless signal propagation. Accurate simulation of wireless networks requires realistic m odels of the channel propag ation medium. An in-depth analysis of the effects of high mobility of MANETs on the performance three prominent ad hoc routing protocols; DS R, AODV (reactive protocols) and DSDV (proactive protocol) under realistic mobility and multi-path fading environments was not performed before. Earlier, the effects of realistic mobility characteristics (temporal dependence of velocity, spatial dependence of velocity, and geogra phic restrictions) on the performance of various ad hoc routing protocols in multi-path fading conditions was not studied in detail. 1.5 Research Objectives This thesis work performs an in-depth st udy on the effects of realistic mobility and fading on energy consumption, packet delivery ratio and control overhead of MANETs. It addresses the issues regarding the effect of realistic mobility characteristics on energy efficiency. Network scenarios which mimic realistic environments are used and the effects of various mobility models on prot ocol performance are to be observed. The simulations are carried out using ns-2 (Network Simulator). The energy consumed by the data (CBR C onstant Bit Rate) and control packets using IEEE 802.11 MAC protocol with the various r ealistic mobility models in multi-path fading channel conditions are obtained. These results are also compared with those obtained for random waypoint mobility model . We also generate snapshots of the node movement in the various mobility mode ls so that a lucid understanding of their
14 characteristics is possible. We evaluate the effects of realistic mobility and fading environment on scalability based on protoc ol performance metrics such as Packet Delivery Ratio (PDR) and Control Overhead. 1.6 Thesis Organization The rest of the thesis is organized as follo ws. Next Chapter briefly describes the related literature review and bac kground of the fading channe l, IEEE 802.11 MAC, ad hoc routing protocols and realistic mobility mode ls. Chapter 3 investigates the effects of multi-path fading and realistic mobility characteristics temporal dependency, spatial dependency and geographic restricti on on the energy utilization, throughput performance, scalability and packet drop rate of MANETs. Chapter 4 describes our network environment, simulation settings and mobility generators. In Chapter 5 the simulation results are presented and in Ch apter 6, we present our conclusions and discuss future work. 1.7 Summary In this chapter, we briefly studied wirele ss LANs and mobile ad hoc networks. The main challenges faced by MANETs both in research and real environments were discussed. Next we presented the motivation for the thesis and an overview of how it is organized.
15 Chapter Two Literature Review and Background 2.1 Literature Review Reference  investigated th e impact of wireless fading ch annel models on the accuracy and evaluation time of large-scale simulation models. In , Bernard Sklar addresses Rayleigh fading, primarily in the UHF band, which affects mobile systems. It also studies the fundamental fading manifestations and the types of degradation. A simple method for modeling small scale Ricean (or Rayleigh) fading is introduced in . It also demonstrates a computationally efficient way to model small-scale fading statistics within a packet level simulator. A set of phys ical layer factors such as signal reception, path loss, fading, interference, noise comput ation and preamble length are presented in  to evaluate the performance of ad hoc routing protocols such as DSR and AODV. In , Hong Jiang et al. analyzed the performance of three routing protocols AODV, DSR and STAR compared in terms of cont rol overhead, amount of data delivered and average latency in packet delivery. Refere nce  focuses on the energy consumption and studies the range effects of DS R, AODV and DSDV and how changes to transmission power and transmission radius affect the overall energy consumed by routing related packets using the random waypoint mobility model. But in  JuanCarlos Cano et al. measured and compared energy consumption behavior of DSR, AODV, DSDV and TORA by varying pause time, maximum node speed, number of traffic sources, number of nodes, simulation area and sending rate. Amit Jardosh et al. proposed to create a more real istic movement model through the incorporation of obstacles wh ich are used to restrict both node movement as well as wireless transmissions [ 34]. In , Bor-rong Chen et al. performed an energy-based comparison of AODV, DSR, DSDV and TORA using three different mobility models:
16 RW model, RPGM model and MG model. Th ey showed significant energy conservation performance difference among mobility models. Fan Bai et al. developed a framework called IMPORTANT ( I mpact of M obility on P erformance Of R ou T ing protocols for A d hoc N e T works)  to evaluate the impact of different mobility models such as random waypoint (RW) model, reference point group mobility (RPGM) model, freeway model and Manhattan grid (MG) model on the performa nce of popular MANET routing protocols (DSR, AODV and DS DV) and also proposed vari ous protocol independent metrics such as spatial dependence, temporal dependence and geographic restrictions to capture interesting mobility characteristics. 2.2 Fading Channel Fading is a variation of signal power at receivers caused by the node mobility or environmental changes that create varying pr opagation conditions from transmitters . There are three main mechanisms that impact radio propagation in wireless channels: Reflection, Diffraction and Scatte ring as shown in Figure 2.1. Reflection occurs when an electromagnetic wave impinges on a smooth surface with very large dimensions when compared to the wave length of the radio wave. It may interfere constructively or destructively at the receiver. Diffraction occurs when the path of the electromagnetic wavefront is obstructed a nd deviated by an impenetrable body of large dimensions as compared to the RF signa l wavelength. Diffraction is also called shadowing because the diffracted field can re ach the receiver even when shadowed by an impenetrable obstruction. Scattering occurs when the radio channel contains objects of dimensions that are on the or der (or less) of the electr omagnetic wavelength, causing energy from a transmitter to be radiated in many different directions. Scattering results in a disordered or random change in th e incident energy distribution .
17 Figure 2.1. Illustration of Fading Mechanisms There are two major categories of fading: Large Scale Fading, Small Scale Fading. Large Scale Fading is the loss that most propag ation models try to account for. They are mostly dependant on the distance from the tran smitter to the receiver. It is also known as Large Scale Path Loss, Log-Normal Fadi ng, or Shadowing. Small Scale Fading is caused by the superposition or cancellation of multipath propagation signals, the speed of the transmitter or receiver, and the bandwidth of the transmitted signal as illustrated in Figure 2.2. It is also known as Multipath Fading, Rayleigh Fading, or simply as Fading. Figure 2.2. Multi-path or Rayleigh Fading
18 MANET scenarios undergo fading which presen ts Rayleigh or Ricean distributions, depending on the geometrical conditions . Rayleigh fading is the fading in a channel due to the interference caused between the direct signal an d the same signal traveling over different paths, resulting in out-of-phase components incident at the receiver . The fading with the Rayleigh distribution is used for mobiles with no line of sight (NLOS) between the transmitter and the receiv er. Rayleigh fading with strong line of sight content is said to be Ricean fading. The signal level from the Ricean path with respect to the power from Rayleigh paths can be controlled by a parameter called Ricean K factor . The additive white Gaussian noise (AWGN) model is used to model an idealistic channel condition wh ere no signal fading occurs. 2.3 IEEE 802.11 Medium Access Control (MAC) IEEE 802.11 is the most widely adopted prot ocol standard for wireless local area networks (WLANs). It specifies two different modes: the infrastructure mode and the ad hoc mode. A special device, Access Point (AP), must be presented as the central point of each Basic Service Set (BSS) in the infrastr ucture mode. Communications inside a BSS happen only between AP and stations, and AP usually connects to the wireline network as the gateway to the Internet. The architecture of infrastructure mode is like a cellular network where a Base Station is the center of each cell. In the ad hoc mode, current standards are built on an environment wher e stations in a grouup are all within each others transmission range, and they communi cate in a peer-topeer fashion. In other words, the ad hoc mode of 802.11 supports only single-hop ad hoc networks, referred to in the specification as Independent Basic Service Sets (IBSSs) . IEEE 802.11 Architecture: IEEE 802.11 networks ar e comprised of Stations, Wireless Medium, AP (Access Points) and a DS (Distr ibution System), as shown in Figure 2.3.
19 Figure 2.3 Schematic of IEEE 802.11 Network Architecture  Station : It is any device which as a I EEE 802.11 MAC (Medium Access Control) and Physical layer interface to the wireless medi um. It maybe a laptop computer or a PDA (Personal Digital Assistant). Access Point (AP) : It is a device found within an IEEE 802.11 netw ork, which provides the point of interconnection between the wireless station and wired network. There are various types of access points and base stations used in both wireless and wired networks. These include bridges, hubs, switches, rout ers and gateways. The differences between them are not always precise, because certain capabilities associated with one can also be added to another. Distribution System (DS) : A DS is a logical element of IEEE 802.11 network that provides a means of connecting multiple APs together . IEEE 802.11 MAC mainly relies on two techniques to combat interference: physical carrier sensing and RTS/CTS (request-to-send / clear-to-send) handshake (also known as virtual carrier sensing). Ideally, th e RTS/CTS handshake can eliminate most interference. However, the effectivene ss of RTS/CTS handshake is based on the assumption that hidden nodes are within tran smission range of receivers. Resolving hidden terminal problem becomes one of the major design considerations of MAC protocols. IEEE 802.11 DCF (Distributed Cont rol Function) is the most popular MAC protocol used in both wireless LANs and m obile ad hoc networks (MANETs) . This protocol generally follows the CSMA/CA (Carrier Sense Multiple Access / Collision
20 Avoidance) paradigm, with extensions to allow for the exchange of RTS-CTS (requestto-send/clear-to-send) handshake packets betw een the transmitter and the receiver. These control packets are needed to reserve a transmission floor for the subsequent data packets. Nodes transmit their control and data packets at a common maximum power level, preventing all other potentially in terfering nodes from starting their own transmissions. Any node that hears the RTS or the CTS message defers its transmission until the ongoing transmission is over. While su ch an approach is fundamentally needed to avoid the hidden terminal problem, it nega tively impacts the channel utilization by not allowing concurrent transmissions to ta ke place over the rese rved floor . 2.4 Routing Protocols Routing protocols are categorized as proactive and reactive protocols. Proactive routing protocols DSDV (Destination Sequenced Dist ance Vector Routing) are table-driven protocols; they always maintain current upto-date routing information by sending control messages periodically between the hosts whic h update their routing tables. Reactive or on-demand routing protocols are the ones which create routes when they are needed by the source host and these routes are mainta ined while they are needed. DSR (Dynamic Source Routing) and AODV (Ad hoc On-demand Distance Vector Rou ting) are the most popular reactive routing protocols. 2.4.1 Dynamic Source Routing (DSR) Protocol The Dynamic Source Routing (DSR)  is an on-demand routing protocol that is based on the concept of source routing. Mobile nodes are required to maintain route caches that contain the source routes of which the mobile is aware. Entries in the route cache are continually updated as new routes are learne d. The protocol c onsists of two major phases: a) route discovery, and b) route maintenance When a source wishes to communicate with a destination, a source st arts with a route discovery by flooding a route request packet. The route request message contains the address of the destination along with the source nodes address and a unique identification number. Each node receiving the packet checks wh ether it knows of a route to th e destination, if not, it adds
21 its own address to the route record of the packet along its outgoing links. A node discards the route request, if it finds its own a ddress already recorded in the route. A route reply is generated when either the route re quest reaches the destination itself, or when it reaches an intermediate node that cont ains in its cache an unexpired route to the destination. When the destination receives a request packet, it may simply reverse the recorded route to reach the source or may use the sa me route discovery procedure toward the original source. Route maintenance is accomplished through the use of route error packets and acknowledgements The route error packets are generated at a node when the data link layer encounters a fatal transmissi on problem. When a route error packet is received, the hop in error is removed from the nodes route cache and all the routes containing the hop are truncated at that point. In addition to route error messages, acknowledgements are used to verify th e correct operation of the route links. 2.4.2 Ad hoc On-Demand Distance Vector (AODV) Routing Protocol The Ad hoc On-demand Distance Vector ( AODV) routing protocol  builds on the DSDV (Destination Sequenced Distance Vector) algorithm It is an improvement on DSDV because it typically minimizes the number of required broadcasts by creating routes on an on-demand basis, as opposed to maintaining a complete list of routes as in the DSDV algorithm. It is classified as a pure on-demand route acquisition system When a source wishes to send a message to so me destination and doe s not already have a valid route to that destination, it initiates a route discovery process to locate the other node. It broadcasts a route re quest (RREQ) packet to its neighbors, which then forward the request to their neighbors, and so on, until either the destination or an intermediate node with a recent route to the destination is located. AODV uses destination sequence numbers to ensure that all routes are loopfree and contain the most recent route information. Route maintenance is carried out by the use of link failure notification message (an RREP with an infinite metric) which is propagated by upstream neighbors (which notice a nodes movement) to each of its active upstream neighbors to inform them of the erasure of that part of the route. AODV additionally uses hello messages which are periodic local broadcasts made by a node to inform each mobile node of other
22 nodes in its neighborhood. Hello message s can be used to maintain the local connectivity of a node 2.4.3 Destination-Sequenced Distance-V ector (DSDV) Routing Protocol The Destination-Sequenced Distance-Vector (D SDV) routing protocol is a table-driven protocol requiring every node to periodically propagate routing information updates throughout the network . Each node periodically broadcasts its rou ting table to all of its neighbor nodes and this route informati on will be propagated from the source node through the network until it re aches the destination node. Route maintenance in DSDV is different from that in DSR and AODV. Whenever signifi cant changes of topology happen, e.g. a MN (Mobile Node) detects a break in the link or it discovers a new neighbor in its proximity, MNs will broadcast its routing tabl e. Each node receiving this information should also broadcast the topology update to its neighbor s . In DSDV, each node maintains a routing table indexed by seque nce numbers, and listing the next hop for every reachable destination. The seque nce numbers enable the mobile nodes to distinguish stale routes from new ones. To maintain table consistency each node periodically transmits the r outing table over the network. 2.5 Mobility Models 2.5.1 Random Waypoint Model (RW) It is a benchmark model to evaluate th e MANET routing protocols, because of its simplicity and wide availability. This mobility model includes pause times between changes in direction and/or speed. An MN (Mobile Node) begins by remaining in a particular location for a certain pause time. Once this time expires, the MN chooses a random destination in the network area and a sp eed that is uniforml y distributed between [minspeed, maxspeed] and travels toward th e new destination, as shown in Figure 2.4. Upon arrival, the MN pauses for a specified time period before repeating the process again. To generate the node trace of the RW model the setdest tool from the CMU Monarch group is used which is included in ns2  .
23 NODE MOVEMENT USING RANDOM WAYPOINT MOBILITY MODEL0 50 100 150 200 250 300 350 400 450 500 050100150200250300350400450500X VALUES (meters)Y VALUES (meters) RANDOM WAYPOINT Figure 2.4. Traveling Pattern of a Mobile No de using Random Waypoint Mobility Model 2.5.2 Gauss Markov Model (GM) In this model, the velocity of mobile node is assumed to be correlated over time and modeled as a Gauss-Markov stochastic process. It is a temporally dependent mobility model where the degree of dependency is determined by the memory level parameter By tuning this parameter various scenarios are obtained: (i) = 0 then the model is memoryless, (ii) = 1 then the model has strong memory and (iii) 0 < <1 then the model has some memory . Figure 2.5. illustrates the trav eling pattern of an MN us ing the GM mobility model. This mobility model can eliminate the sudden stops and sharp turns encountered in the random walk mobility model by allowing past velocitie s to influence future velocities (i.e) introducing temporal dependency of velocity.
24 NODE MOVEMENT USING GAUSS MARKOV MOBILITY MODEL0 100 200 300 400 500 600 0100200300400500600X VALUES (meters)Y VALUES (meters) Gauss Markov Figure 2.5. Traveling Pattern of a Mobile Node using Gauss Markov Mobility Model 2.5.3 Manhattan Grid Model (MG) This model emulates the movement pattern of mobile nodes on streets defined by maps. It is useful in modeling movement in an urba n area. Maps are used in this model which is composed of a number of horiz ontal and vertic al streets. NODE MOVEMENT USING MANHATTAN GRID MOBILITY MODEL0 50 100 150 200 250 300 350 400 450 500 0100200300400500 X VALUES (meters)Y VALUES (meters) MANHATTAN GRID Figure 2.6. Traveling Pattern of a Mobile No de using Manhattan Grid Mobility Model
25 The mobile node is allowed to move along the grid of horizontal a nd vertical streets on the map. At an intersection the MN can turn right, left or go straight as depicted in Figure 2.6. This model has high temporal dependency of velocity as well as spatial dependency of velocity. Also it imposes geographic rest rictions on the moveme nt of the MN  . 2.5.4 Reference Point Group Mobility (RPGM) Model In an ad hoc network there are a lot of s cenarios where it is necessary to model the behavior of MNs as they move together Group mobility can be used in military battlefield communication, rescue operations, tracking etc . Each group has a logical center (group leader) that determines the groups motion behavior. Initially, each member of the group is uniformly distributed in the neighborhood of the group leader. Subsequently, at each instan t, every node has a speed and direction that is derived by randomly deviating from that of the group le ader. Figure 2.7. illust rates the traveling pattern of three MNs moving together as one group. Figure 2.7. Traveling Patterns of Th ree Mobile Nodes using Reference Point Group Mobility Model
26 The movement of the logical center for each group, and the random motion of each individual MN within the group, are implemen ted via the RW Mobility Model. However, the individual MNs do not use pause times while the group is moving. 2.5.5 Pursue Mobility Model (PM) It emulates scenarios where several nodes attempt to capture a single mobile node ahead. This mobility model could be used in targ et tracking and law enforcement. The node being pursued moves freely according to the RW model. The current position of an MN, a random vector, and an acceleration function ar e combined to calculate the next position of the MN. By directing the velocity towa rds the position of th e targeted node, the pursuer nodes try to inte rcept the target node as seen in Figure 2.8 . Figure 2.8. Movement of Mobile N odes using Pursue Mobility Model 2.5.6 Column Mobility Model (CM) The column mobility model represents a se t of mobile nodes (e.g., robots) that move in a certain fixed direction. This mobility model can be used in searching and scanning activity, such as destroying mines by military robots . This model represents a set of MNs that move around a given line, which is moving in a forward direction. For the
27 implementation, an initial reference grid is de fined as shown in Figure 2.9. Each MN is then placed in relation to its reference point in the reference grid; the MN is then allowed to move randomly around its reference point via an entity mobility model such as RW model or random walk model. Figure 2.9. Movement of Mobile Nodes using Column Mobility Model 2.6 Summary In this chapter, we reviewed some of the pr evious research done in evaluating the effect of multi-fading, energy consumption and m obility on the performance of routing protocols in MANETs. We then briefly de scribed the fading channel, IEEE 802.11 MAC, routing protocols and mobility models that are considered in our network simulations.
28 Chapter Three Factors Influencing MANET Performa nce in Realistic Environments 3.1 Network Simulations in Realistic Environments Mobile ad hoc network performance can be evaluated through th e use of simulation as it provides the capability to analyze the effect of different protocol parameters on different performance metrics in various netw ork scenarios. Previously, most of the simulations of MANETs were done using RW mobility model as a default model. In this entity model, the Mobile Node (MN) moves in a random fashion with a specific pause time. But the scenarios in which ad hoc networks are implemented, the node mobility may not be randomized. Hence it is necessary to evaluate MANET performance with traveli ng patterns that emulate human movements. Figure 3.1 shows the network simulation settings of a MANET in an unrealistic environment incorporated with RW mobility model, two ray ground reflection radio model and CBR traffic. Two ray ground propagation model considers the direct path and the ground reflection path when calculat ing the received signal power of each packet. Though it is more accurate than the free space model, which assumes the ideal propagation condition that there is only one line-of-sight path between the transmitter and receiver , this model does not take into consideration the effect of multi-path fading on the wireless channel. Figure 3.2 illustrates the network simulation envir onment created for our MANET performance evaluation. We incorporated mobility models with realistic movement characteristics such as temporal depende nce of velocity, spatial dependence of velocity and geographic restrictions. The m obility models are broadly categorized as Entity and Group mobility models . En tity mobility model specifies individual
29 node movement. Group mobility model describes group movement as well as individual node movement inside groups. The entity models considered were GM Model and MG Model. The group mobility presented was RPGM model, CM model and PM model. Table 3.1 tabulates the mobility models used for our simulations and the realistic mobility character istics they exhibit. Figure 3.1. Network Simulation in an Unrealistic Environment 3.2 Temporal Dependency of Velocity Temporal dependence of velocity indicates the similarity in the velocities of a node within a specified time interval . In most real life scenarios, the speed of vehicles and pedestrians will accelerate incrementa lly. The direction change will also be smooth. Hence the velocity at current time period is depende nt on the previous epoch, i.e., the velocities of a node at different time slots are correlated  . So the mobility model should have some memory to prevent extreme mobility behavior, such as sudden stop, sudden acceleration a nd sharp turn, which may frequently occur
30 in the trace generated by the RW model. Hence we use GM model so that the role of temporal dependence of velocity on MANET performance can be understood. Figure 3.2. Network Simulation in a Realistic Environment 3.3 Spatial Dependency of Velocity Spatial dependence of velocity indicates th e similarity in the velocities of two nodes that are within a specified transmission ra nge from each other, i.e., the velocities of different nodes are correlated in space . In some scenarios such as battlefield communication and museum touring, the move ment pattern of a mobile node may be influenced by a certain group leader node in its neighborhood  . Hence, the mobility of various nodes is indeed correla ted. But the RW model considers a mobile node as an entity that moves independently of other nodes. So we need to consider
31 using group mobility models like RPGM model, CM model and PM model that characterizes inter-depende nt movement of nodes. 3.4 Geographic Restriction of Movements RW and its variants assume that the m obile nodes can move freely within the simulation field without any restrictions. Howe ver, in realistic applications in urban area settings, the movement of a mobile node may be bounded by obstacles, buildings, streets or freeways  . As the nodes m ovement is subject to the physical conditions they will move in a ps eudo-random fashion on a predefined path. Some realistic mobility models incorporate the predefined paths and obstacles into the mobility models. We use MG mobility model in our simulations which restricts the movement of the mobile node to the pathway in the simulation field. Figure 3.3. Pathway Map of Manha ttan Grid Mobility Model In mobility models with geogr aphic restrictions, the predefined pathways restrict and partly define the movement path of nodes even though there exists a certain level of randomness. Hence the pathway of the simulation field is a key element for characterizing the geographic constraint of a mobility model. The pathway map used for MG mobility model is illustrated in Figure 3.3.
32 Table 3.1 Mobility Models and th eir Movement Characteristics MOBILITY MODELS TEMPORAL DEPENDENCE OF VELOCITY SPATIAL DEPENDENCE OF VELOCITY GEOGRAPHIC RESTRICTIONS/ OBSTACLES Random Waypoint No No No Manhattan Grid Yes No Yes Gauss Markov Yes No No RPGM No Yes No Column Motion No Yes No Pursue Motion No Yes No 3.5 Effect of Multi-path Fading Channel Figure 3.4 illustrates the effect of Rayl eigh fading on a signals envelope. The time interval corresponding to two adjacent smallscale fades is on the order of a half wavelength ( / 2). Figure 3.4. Time-Se quenced Rayleigh Fading Envelope
33 3.6 Scalability of a Network As wireless ad hoc networks used for security and commercial purposes have significantly magnified in their sizes over th e years, it is necessary to explore the behavior of large scale, complex wirele ss network systems in realistic ad hoc environments. Thus there is a need to use high performance simulation tools to achieve scalability to large networks  But due to the limitations in the network simulator (Ns-2) used for our simulati ons the number of nodes in our network designed is around 50. Performance metrics such as packet delivery ratio (PDR) and control overhead are employed to analyze the scal ability of the mobile ad hoc network when simulated in a fading envi ronment with realistic mobility models. 3.7 Summary This chapter reviews the need for realistic ad hoc scenarios in simulations, some of the important realistic movement characteri stics and the mobility models that were used in our simulations. It explains th e effect of Rayleigh fading on a signal. The scalability of MANETs is also discussed briefly.
34 Chapter Four Network Simulation Environment 4.1 ns-2 (Network Simulator) 4.1.1 Origins The simulations are carried out using ns-2 (26th release). ns-2 is a discrete event, object oriented, simulator developed by th e VINT project research group at the University of California at Berkeley targeted at networking research. ns-2 provides substantial support for simula tion of TCP, routing, and mu lticast protocols over wired and wireless (local and satellite) networks. ns-2 began as a variant of the REAL network simulator in 1989 and has evolved substantially over the past few years. In 1995 ns-2 development was supported by DARPA through the VINT project at LBL, Xerox PARC, UCB, and USC/ISI. Currently Ns development is supported through DARPA with SAMAN and through NSF with CONSER both in collaboration with other researchers including ACIRI ns-2 has always included substantial contribu tions from other res earchers, including wireless code from the UCB Daedelus and CMU (Carnegie Mellon University) Monarch projects and Sun Microsystems. The simulator has been extended by the Monarch research group at CMU to include: nodes mobility, a realistic physical layer that includes a radio propagation m odel, and the IEEE 802.11 Medium Access Control (MAC) protocol . 4.1.2 Functional Description ns-2 is a simulator, written in C++ with an OTcl (Object Tool Command Language) interpreter as a front-end. C++ is used for detailed protocol implementation which efficiently manipulates bytes, packet headers, and implements algorithms that run
35 over large data sets. On the other hand OTcl is ideal for slightly varying parameters and simulation configurations, or quickly e xploring a number of scenarios. One of the main advantages of the split -language implementation of ns-2 is its object oriented design, which allows for easy replacement of the software modules involved in a simulation for example a routing protocol a network applica tion, or a propagation model. The process of configuring the set of modules required to perform a particular simulation, starting from the phys ical interface model up to the application layer, is known as plumbing and is usually performed by an OTcl script. When testing a new protocol, or implementing a simulation model, we need to write the code with the correct bindings to the OTcl interface, and afterwards instruct th e plumbing script to employ the newly created modul es during simulation setup. MobileNode is the basic ns Node object with added f unctionalities like movement, ability to transmit and receive on a channel that allows it to be used to create mobile, wireless simulation environments. The clas s MobileNode is derived from the base class Node. MobileNode is a split object. The mobility features including node movement, periodic position updates, maintaining topology boundary etc are implemented in C++ while plumbing of network components within MobileNode itself (like classifiers, dmux LL, Mac, Chan nel etc) have been implemented in Otcl . Figure 4.1 illustrates the plumbing for the network stack objects of a MANET node that uses the DSDV routing protocol: an application layer module, the routing protocol, the address resolution protocol (ARP) module, a link layer (LL) object, an interface queue, the MAC protocol, and th e physical interface with the channels radio propagation model .
36 Figure 4.1. Schematic of a MobileNode under the CMU Monarch's Wireless Extensions to ns-2  4.1.3 Modifications to ns-2 The Ricean (or Rayleigh) propagation model with a Ricean K factor of 0 is included so as to incorporate Rayleigh fading in the channel. A dataset containing the components of a time-sequenced fading envelo pe is pre-computed. With a few simple mathematical operations during the simula tion run, this single lookup table can be used to model a wide range of parameters. The parameters to be adjusted are the timeaveraged power, P, the maximum Doppler frequency, fm, and the Ricean K factor. The signal power from the LOS path with respect to the NLOS paths can be
37 controlled by the Ricean K factor. Although the dataset represents a limited length time sequence, long simulations can be performed by using this limited dataset over and over again. The dataset is constructed so that there are no discontinuities when the sequence repeats. It is assumed that the small scale fading envelope is used to modulate the calculation s of a large scale propagation model (two-ray ground or some other deterministic model) . The Rayleigh (Multipath) Fading modeled in our simulations has the values as shown in Table 4.1: Table 4.1. Multipath Fading Model Parameters PARAMETER VALUES Distribution Ricean Gaussian components Fm 20 Hz N 15584 Fs 1000 Hz Ricean K Factor 0 MaxVelocity 2.5 / 5.0 LoadRiceFile rice_table.txt 4.2 Mobility Generators 4.2.1 The setdest Mobility Generator The RW model is most commonly used mobility model in research of MANETs. This model is provided by the setdest tool in the standard ns-2 distribution . Usage : The syntax  to run setdest with arguments is as shown below: Syntax : ./setdest [-n num_of_nodes] [-p pausetim e] [-s maxspeed] [-t simtime] [-x maxx] [-y maxy] > [ outdir/movement-file]
38 4.2.2 BonnMotion Mobility Generator BonnMotion is Java-based software which cr eates and analyses mobility scenarios. It is developed within the Communication Syst ems group at the Institute of Computer Science IV of the Univers ity of Bonn, Germany. It se rves as a tool for the investigation of mobile ad hoc network characteristics. The scenarios generated in this mobility generator can be exported for ns-2 or GloMoSim. The mobility models that are supported are RW model, GM model, MG model and RPGM model . Table 4.2. Parameters used to Generate Manhattan Grid Movement Pattern GAUSS MARKOV MOBILITY MODEL: Usage : All applications described above are started via the "b m" wrapper script . Syntax : ./bm
39 Here, the parameters for simulating the mobility models are those described in the Tables 4.2, 4.3 below and the application can be a mobility model or e.g. the Statistics application used to analyze th e scenario characteristics. We generate MG movement files for five different speeds: 5 m/s, 20 m/s, 40 m/s, 60 m/s and 80 m/s and export them so that it can be used for simulations in ns-2 High speeds of around 80 m/s are reasonable whenever a MANET includes highly mobile nodes such as helicopters, police, military and other emergency vehicles. Table 4.3. Parameters used to Genera te Gauss Markov Movement Pattern GM movement files for five different speed s: 5 m/s, 20 m/s, 40 m/s, 60 m/s and 80 m/s were generated and then exported them to be used for simulations in ns-2 The angle standard deviation can be varied between 0 and 1. PARAMETER VALUES Model Gauss Markov Mobility Generator BonnMotion Number of Nodes 49 Simulation Time 600 X Dimension 500 Y Dimension 500 Random Seed 1 Angle Standard Deviation 0 Maximum Pause Time 0.1 Pause Probability 0.1 Maximum Speed Varied as speed varies from 5 m/s to 80 m/s
40 4.2.3 Scenario Generator It is a tool to generate MANET mobility scenarios for ns-2 The mobility models that have been implemented include RW model, PM model, GM model and CM model. Also hybrid models can be constructed so th at realistic ad hoc situations such as disaster, conference, etc. can be implemented in simulations . We use this mobility model to generate PM Model for a group of 50 mobile nodes. The movement patterns for five different speeds: 5 m/s, 20 m/s, 40 m/s, 60 m/s and 80 m/s was generated. Usage : The syntax used is as shown, Syntax : ./scengen > outdir/movement-file. The script takes two inputs: "model-spec", wh ich contains the default parameters, and normally does not need to be changed and the other is the scenario specification file "scen-spec", which describes the scenar io needed. The scen-spec for pursue mobility model was generated and the movement file was obtained. The nodes pursuing the runaway node have a direction th at at any instant of time will be in a straight line towards the runaway node. The parameters which model the PM model used in our simulations is as shown in Table 4.4. 4.2.4 Mobility Generator Toilers Code Toilers is an ad hoc res earch group at Colorado School of mines . We use the mobility model codes developed by this group to generate the movement patterns for RPGM and CM models.
41 Table 4.4. Parameters used to Generate Pursue Movement Pattern PARAMETER VALUES Model Pursue Motion Mobility Generator Scenario Generator Start time 0 Seconds Stop time 600 Seconds Number of Nodes 50 (one node being pursued by the rest) X Dimension 500 Y Dimension 500 Maximum Speed of Leader (Pursued) Node Varied as 10 m/s, 25 m/s, 45 m/s, 65 m/s and 85 m/s. Maximum Speed of other (Pursuer) Node Varied as 5 m/s, 20 m/s, 40 m/s, 60 m/s and 80 m/s. Minimum Speed of Leader (Pursued) Node 5 m/s Minimum Speed of other (Pursuer) Node 0 m/s Usage (For RPGM) : ./rpgm
42 keeping the number of nodes in the networ k scenario constant at 50. The different scenario movement files genera ted are as shown in Table 4.6. Table 4.5. Parameters used to Genera te Gauss Markov Movement Pattern The parameters used for CM model are similar to those described for RPGM. Usage (For Column-Line) : ./col-line
43 Table 4.6. Different RPGM Scenarios Simulated GROUP SCENARIO NUMBER OF GROUPS NUMBER OF NODES PER GROUP RPGM1 1 50 RPGM2 2 25 RPGM3 5 10 RPGM4 10 5 RPGM5 25 2 4.3 Traffic Generation We generate 12 Constant Bit Rate (CBR) tr affic connections with send rate of 4 and packet size of 512 bytes for UDP sources. The source-destination pairs are spread over the network as shown in Figure 4.2. Random traffic connections of CBR can be setup between mobile nodes using a traffic-scenario generator. This script is available in ns-2 . It can be used to create CBR and TCP traffics connections between wireless mobile nodes. Syntax : ns cbrgen.tcl [-type cbr] [-nn nodes] [-seed seed] [-mc connections] [-rate rate] 4.4 Network Scenario Two different scenarios are considered base d on the mobility models used. For the Entity Mobility models like RW model, MG model and GM model we use the Scenario A. For Group Mobility models like RPGM model, PM model and CM model we conduct experiments using the Scenario B. Scenario A : We generated an ad hoc network with 49 highly mobile nodes. The simulation area is 500 m x 500m and the simulation time was set to 600 seconds.
44 Figure 4.2 shows the initial position of the nodes and the connections through which the traffic flows. Figure 4.2. Network Schematic S howing Initial Node Positions Scenario B : This scenario models the mobile nodes in groups. Based on the number of groups generated, we consider five di fferent group scenario cases as shown in Table 4.7. Figure 4.3 illustrates the group scen ario 4 where five groups were formed with ten mobile nodes in each group. Table 4.7. Different Gr oup Scenarios Simulated SCENARIO NUMBER OF GR OUPS NUMBER OF NODES PER GROUP Group Scenario 1 1 50 Group Scenario 2 2 25 Group Scenario 3 5 10 Group Scenario 4 10 5 Group Scenario 5 25 2 Source Sink
45 The simulation area is set to be 500 m x 500 m. Simulati ons are run for 600 seconds for 50 nodes. Each data point represents an av erage of at least five runs with identical traffic models, but different random ly generated mobility scenarios. Figure 4.3. Illustration of Group Scenario 3 4.5 Energy Consumption Model According to the specificat ion of the Network Interf ace Card (NIC) modeled, the energy consumption varies from 230mA in receiving mode to 330mA in transmitting mode, using 3.3V or 5.0V voltage supply . All nodes are equipped with IEEE 802.11 NICs with data rates of 2 Mbps. The energy expenditure needed to transmit / receive a packet p is: E(p tx / rcv) = i v* t p Joules, where i is the current value, v the voltage, and t p the time taken to transmit / receive the packet p. Packet transmission time, t p = (packet-size in b its / 2 106) sec. In our simulations, the measured values of a Cabletron Roamabout 802.11 DS high rate NIC operating in base station mode is used. Table 4.8 shows the power cons umption values of the four modes: Transmit mode, Receive mo de, Idle mode and Sleep mode.
46 Table 4.8. Power Consumption Values Transmit Mode 1400 mW Receive Mode 1000 mW Idle Mode 830 mW Sleep Mode 130 mW 4.6 Performance Evaluation The following performance metrics ar e considered in our simulations: Energy-Goodput: It is defined as the ratio of the total bits transmitted to the total energy consumed, where the total bits transmitted are calculated for application layer data packets only and the total energy consumed captures the entire energy utilization of the network with all the control overhead in cluded. The unit for ener gy-goodput is bits/J. Packet Delivery Ratio (PDR) : PDR is the ratio between the number of packets received by the end-point application a nd the number of packets orig inated at the source-node application. Packet Overhead (Control Overhead): Packet Overhead is the number of per-hop nondata packets in the network per originating data packet. In case of CBR applications, this is directly proportional to the number of per-hop rou ting packets in the network. 4.7 Summary This chapter focused on the network simula tion environment used for our research. It also briefly explained the mobility and traffi c generators used to generate movement and traffic patterns. Finally the performan ce metrics that were used to determine protocol performance was presented.
47 Chapter Five Experimental Results 5.1 Scalability in Mobile Ad hoc Networks Scalability of a protocol can be obtained by measuring the protocol performance in different scalable scenarios. Traditionally ad hoc netw orks have been used under small and low-density environments. Larg e-scale mobile ad hoc networks are characterized by high node density, high mobility and large number of nodes. Protocol performances we re evaluated based on Packet Delivery Ratio (PDR) and Control Overhead. As mentioned earlier, two sets of analysis are made to evaluate the performance of MANETs with entity mobility models and group mobility models. In  scalability analysis for MANETs was performed us ing DSR and DSDV considering only Random Waypoint Mobility model. Our simu lations encompass scalibility analysis for some of the most popular entity and gr oup mobility models cited in Chapter 2. 5.1.1 Scalability Analysis using Entity Mobility Models The Random Waypoint (RW) model is the de fault model which does not include any of the realistic mobility characteristics mentioned in Chapter 3. Manhattan Grid model has geographic restrictions incorpor ated in it through the use of pathway graphs. It was initially exp ected to have high degree of spatial dependency as the mobility of a node is subjective to the moveme nt of the nodes ahead of it, in the lane. But from  we understand that Manhattan Grid (MG) model has negligible spatial dependence of velocity as the positive de gree of spatial depe ndence (due to nodes traveling in same direction) is cancel led out by the negative degree of spatial dependence (due to nodes traveling in opposite direction). Gauss Markov (GM)
48 mobility model has a high degree of temporal dependence as the velocity of the node is correlated over time and modeled as a Gauss Markov stochastic process . The maximum speed is increased from 5 m/s to 80 m/s. We keep the network density constant for all our simulati ons and hence any changes in protocol performance can be directly attributed to the mobility m odel used and the variation in speed. Packet delivery ratio is strongly influe nced by the number of packets that are dropped, either at the source nodes or at intermediate nodes. Most packets being dropped are at the intermediate nodes which are mainly due to network congestion or broken links. Figure 5.1. PDR Analysis for Various R outing Algorithms using Random Waypoint Mobility Model
49 Figure 5.2. PDR Analysis for Various Rou ting Algorithms using Manhattan Grid Mobility Model Figure 5.3. PDR Analysis for Various R outing Algorithms using Gauss Markov Mobility Model
50 From Figures 5.1, 5.2 and 5.3 we observed th at DSR has a higher PDR with RW model than AODV and DSDV. But its performance de grades significantly with MG and GM models in comparison with AODV and DSDV One main reason for this performance drop in DSR can be attributed to the fact that control overh ead increases more drastically as speed increases from 5 m/s to 20 m/s, when MG or GM models were used instead of RW model. Higher control overh ead is needed to repair the more frequently occurring link breakages. Surprisingly, from Figures 5.4, 5.5 and 5.6 we observe that the control overhead produced by MG model and GM mo del on AODV and DSDV is lesser when compared to that produced by using RW model. Figure 5.4. Control Overhead Analysis fo r various routing algorithms using Random Waypoint Mobility Model Comparing the PDR analysis made for RW and MG mobility models, we can conclude that when there are geographic restrictions associated with the movement of a mobile node in a MANET, there are more link breakages and hence there are more packets being dropped by forwarding nodes. When more packets are dropped more retransmissions takes place and there is highe r network congestion leading to lower PDR values. As the
51 speed increases the PDR decreases gradually when GM mobility model is used. Here again there are more broken links when compared to RW model. Figure 5.5. Control Overhead Analysis for Various Routing Algorithms using Manhattan Grid Mobility Model Figure 5.6. Control Overhead Analysis for Various Routing Algorithms using Gauss Markov Mobility Model
52 5.1.2 Scalability Analysis using Group Mobility Models The effect of spatial dependence of velocity on the protocol performance in MANETs is a major concern when ad hoc networks are utilized in military operations, rescue missions, tracking and law enforcement, where a group of mobile nodes work and move together to achieve a particular goa l. The scalability of a MANET using group mobility models is analyzed through simulations carried out in ns-2 using the Scenario B mentioned in Chapter 4. It has been proved that the single group m obility has a higher value for degree of spatial dependence than that of multiple group mobility . Hence the degree of spatial dependence of velocity decreases as we go from Group Scenario 1 to Group Scenario 5. These scenarios were explained in detail in Chapter Four. This analysis gives a lucid understanding of the effect of the degree of spatial dependence on the scalability of a protocol when used in a mobile ad hoc network. Figure 5.7. PDR Analysis for DSR using RPGM Mobility model for Various Speeds
53 Figure 5.8. PDR Analysis for DSR using RP GM Mobility Model for Various Group Scenarios Figure 5.9. PDR Analysis for AODV using RPGM Mobility Model for Various Speeds
54 Figure 5.10. PDR Analysis for AODV using RPGM Mobility Model for Various Group Scenarios
55 Figure 5.11. PDR Analysis for DSDV usi ng RPGM Mobility Model for Various Speeds Figure 5.12. PDR Analysis for DSDV usi ng RPGM Mobility Model for Various Group Scenarios The following conclusions can be made from the PDR analysis made when using RPGM mobility model: Considering five different group scenario s, for DSR, as speed increases from 5 m/s to 80 m/s, the PDR decreases from 1.0 to about 0.41. Whereas for AODV the PDR decreases from 1.0 to about 0.54 and for DSDV, the PDR decreases from 1 to 0.726. These results ar e graphically depi cted in Figures 5.7, 5.9 and 5.11. In all cases (DSR, AODV DSDV), the PDR is a ma ximum at 5 m/s for all group scenarios and a minimum for 80 m/s. From Figures 5.8, 5.10 and 5.12 we can infe r that as the degree of spatial dependence of velocity decreases fr om Scenario: RPGM1 to Scenario: RPGM5, PDR decreases in general but DSR shows a more drastic decrease
56 than AODV or DSDV. From the graphs it can be inferred that DSDV shows more consistent values when compared to DSR and AODV. Figure 5.13. Control Overhead Analysis fo r DSR using RPGM Mobility Model for Various Speeds Figure 5.14. Control Overhead Analysis fo r AODV using RPGM Mobility Model for Various Speeds
57 Figure 5.15. Control Overhead Analysis fo r DSDV using RPGM Mobility Model for Various Speeds The following conclusions can be made fr om the control overh ead analysis made when using RPGM mobility model: From Figures 5.13, 5.14 and 5.15 we in fer that the control overhead is maximum for speeds of 80 m/s and minimum for 5 m/s for DSR, AODV and DSDV. These results can be directly related to the PDR results explained earlier. Control overhead is very less in DS R (maximum 43526 control packets) compared to AODV (maximum 60648 control packets) and DSDV (maximum 65755).
58 The control overhead increas es very steeply in AODV as speed increases from 5 m/s to 80 m/s. This is because there is more flooding of route discovery and route request packets as there are more route changes as mobility increases. Figure 5.16. Control Overhead Analysis fo r DSR using RPGM Mobility Model for Various Group Scenarios DSDV is least affected by variation in speed. Because DSDV is a distancevector protocol, it is responsible for periodically announcing its routing table to all one-hop neighbors. Since DSDV r outing tables contain a list of next-hop entries for every node in the ad hoc netw ork, the size of this routing update is independent of a nodes transmission range or power level. DSDV can be expected to be less sensitive to higher mobility rates and hence there is not much change (increase) in the number of link breakages as speed is increased from 5 m/s to 80 m/s. Hence we do not find a drastic increase in the control overhead produced, as depicted in Figure 5.18. In DSR, there is a considerable incr ease in the control overhead produced as the scenario transforms from a singl e group to multiple groups. DSDV shows
59 very slight increase, whereas AODV is the worst affected as the control overhead increases drastica lly as we move from a single group scenario to a multiple scenario. This behavi or is depicted by Table 5.1. Figure 5.17. Control Overhead Analysis fo r AODV using RPGM Mobility Model for Various Group Scenarios Figure 5.18. Control Overhead Analysis fo r DSDV using RPGM Mobility Model for Various Group Scenarios
60 Table 5.1. Maximum Control Overhead Pr oduced for Single and Multiple Group Scenarios when Implementing RPGM Mobility Model ROUTING PROTOCOL MAXIMUM CONTROL-OVERHEAD SINGLE GROUP SCENARIO (Control Packets) MAXIMUM CONTROL-OVERHEAD MULTIPLE GROUP SCENARIO (Control Packets) DSR 2903 40821 AODV 16238 76100 DSDV 61929 63638 Figure 5.19. PDR Analysis for DSR using Colu mn Mobility Model for Various Speeds
61 Figure 5.20. PDR Analysis for DSR using Colu mn Mobility Model for Various Group Scenarios Figure 5.21. PDR Analysis fo r AODV using Column Mobility Model for Various Speeds
62 Figure 5.22. PDR Analysis fo r AODV using Column Mobility Model for Various Group Scenarios Figure 5.23. PDR Analysis for DSDV usi ng Column Mobility Model for Various Speeds
63 Figure 5.24. PDR Analysis for DSDV usi ng Column Mobility Model for Various Group Scenarios The following conclusions can be made from the PDR analysis made when using Column mobility model: Considering five different group scenario s, for DSR, as speed increases from 5 m/s to 80 m/s, the PDR decreases from 1.0 to about 0.4. Whereas for AODV the PDR decreases from 1.0 to about 0.52 and for DSDV, the PDR decreases from 1 to 0.75. These results are gr aphically depicted in Figures 5.16, 5.18 and 5.20. In all cases (DSR, AODV DSDV), the PDR is a ma ximum at 5 m/s for all group scenarios and a minimum for 80 m/s. From Figures 5.17, 5.19 and 5.21 we can in fer that as the de gree of spatial dependence of velocity decreases from Scenario: Column-Line1 to Scenario: Column-Line3, PDR decreases in gene ral but DSR shows a more drastic decrease than AODV or DSDV. From th e graphs it can be inferred that DSDV shows more consistent values wh en compared to DSR and AODV.
64 These results are compared to those obtained with RPGM performance and it is found that they are similar. Column Mobility model can be derived from the RPGM mobility model implementation and this is the underlying reason for such a comparable performance behavior. Figure 5.25. Control Overhead Analysis for DSR using Column Mobility Model for Various Speeds The following conclusions can be made fr om the control overh ead analysis made when using Column mobility model: From Figures 5.25, 5.27 and 5.29 we infer that the control overhead is a maximum for speeds of 80 m/s and a mi nimum for 5 m/s for DSR, AODV and DSDV. These results can be directly related to the PDR results explained earlier. Control overhead is very less in DS R (maximum 43526 control packets) compared to AODV (maximum 60648 control packets) and DSDV (maximum 65755).
65 The control overhead increas es very steeply in AODV as speed increases from 5 m/s to 80 m/s. This is because there is more flooding of route discovery and route request packets as there are more route changes as mobility increases. Figure 5.26. Control Overhead Analysis for DSR using Column Mobility Model for Various Group Scenarios Although DSDV produces the maximum c ontrol overhead with column mobility model, it is least affected when the speed increases from 5 m/s to 80 m/s, as depicted in Figure 5.29. Table 5.2 illustrates the maximum vari ation in control ove rhead produced as the scenario is tr ansformed from a single group to multiple groups. DSR shows a considerable increase in c ontrol overhead and AODV has a drastic increase, whereas DSDV is th e least affected protocol.
66 Table 5.2. Maximum Control Overhead Pr oduced for Single and Multiple Group Scenarios when Implementing Column Mobility Model ROUTING PROTOCOL MAXIMUM CONTROL-OVERHEAD SINGLE GROUP SCENARIO (Control Packets) MAXIMUM CONTROL-OVERHEAD MULTIPLE GROUP SCENARIO (Control Packets) DSR 4415 43526 AODV 25995 60648 DSDV 65755 64230 Figure 5.27. Control Overhead Analysis for AODV using Column Mobility Model for Various Speeds
67 Figure 5.28. Control Overhead Analysis for AODV using Column Mobility Model for Various Group Scenarios Figure 5.29. Control Overhead Analysis fo r DSDV using Column Mobility Model for Various Speeds Pursue mobility model is mostly used in target tracking and law enforcement, where a group of mobile nodes attempt to capture a single node ahead of them. Pursue motion model is also derived from RPGM mobility model.
68 Figure 5.30. Control Overhead Analysis fo r DSDV using Column Mobility Model for Various Group Scenarios From Figure 5.31 it was observed th at the PDR is almost at a value of 1.0 for all three routing algorithms. Though the control overhead remains almost same for all speeds, for all three routing protocols, th e control overhead produced by DSR Figure 5.31. PDR Analysis for Various Routing Algorithms using Pursue Mobility Model
69 Figure 5.32. Control Overhead Analysis for Various Routing Algorithms using Pursue Mobility Model is very less (Maximum = 5038 control packets) when compared to AODV (Maximum = 19618 control packets) and DSDV (M aximum = 69227 control packets). In summary, the following inferences are made from the scalability analysis carried out with group mobility models: RPGM m odel, Column model and Pursue model. In RPGM and Column models, as the ne twork transforms from being a single group scenario to multiple group scenarios, there is lesser homogeneity and therefore there is less route formations between nodes in different groups. This results in longer route formati ons to transmit data across nodes of different groups. In Pursue model, the distribution of nodes is more homogeneous and hence there are shorter route formations between nodes.
70 Although DSR has a high PDR and low control overhead at lower speeds, it has to be noted that there is a drasti c drop in performance as speed increases and hence we need to consider the c onsistency of the protocol performance when the mobile ad hoc network is scal ed to include more number of nodes and operate in high mobility conditions. As speed increases more route changes take place and hence more routing packets are transmitted which in turn has an effect on the p acket delivery of the network. DSR performs better with group mobility scenarios which have high degree of spatial dependence than with those which are less homogeneous (multiple groups). DSDV seems to be more re liable routing protocol to be used in group mobility scenarios with less homogeneity (multiple groups) as there is no severe effect on protocol performance when speed incr eases or when the degree of spatial dependency decreases. As speed increas es, more route changes take place and hence there is a need for more freque nt updates of the routing table which increases routing overhead and hence reduces PDR values. In DSDV, when degree of spatial dependency (homogeneity) decreases, all the route formations within a group ar e usually stable even when mobility increases as the nodes are more closel y packed among themselves and hence these routes need not be updated frequen tly. This helps to maintain the control overhead constant or at the least there is less variation of control overhead. Hence even when the single group transforms to multiple group scenarios, there is very little performance drop. AODV performs similar to DSDV when th e degree of spatial dependence of velocity (homogeneity) decreases, but ther e is a more visible drop in protocol performance than DSDV.
71 5.2 Energy Utilization in Mobile Ad hoc Networks In this section we study the effect of realistic mobility characteristics such as temporal dependence of velocity, spatial dependence and geographic restrictions on the energy utilization in MANETs. Highe r the energy-goodput value better is the energy utilization. 5.2.1 Energy-Goodput Analysis using Entity Mobility Models Random Waypoint model, Manhattan Grid model and Gauss Markov model are the three entity models used. From the energy-goodput anal ysis in Figure 5.33 made the following conclusions can be made: DSR and DSDV perform well in terms of energy with all speeds when Random Waypoint model is used. DSR produces th e least number of routing packets and hence consumes less energy. DSDV on the other hand is less sensitive to variation in speeds, as discussed previously and hen ce the routing packets generated remains almost constant. This contributes to a c onsistent energy-goodput for all speeds from 5 m/s to 80 m/s. Figure 5.33. Energy-Goodput Analysis of Va rious Routing Algorithms using Random Waypoint Mobility Model
72 The energy performance of AODV decreases drastically as speed increases from 5 m/s to 20 m/s. This can be at tributed to the fact that as speed increases the topology changes frequently which in turn cause mo re route changes and hence more routing packets are produced which consumes more power. So the energy-goodput decreases drastically. Figure 5.34. Energy-Goodput Analysis of Various Routing Algorithms using Manhattan Grid Mobility Model Figure 5.34 analyzes the energy perfor mance of DSR, AODV and DSDV using Manhattan Grid mobility model. Comparing this with Figure 5.33 we infer the effect of geographic restrictions on the energy utilization of a mobile node in a MANET. The energy-goodput of DSR decreases steadily from 262 bits/J to 83 bits/J as speed varies from 5 m/s to 80 m/s. This steep fa ll in energy-goodput can be attributed to the geographic restrictions such as predefined pathways which are incorporated into the Manhattan Grid mobility model. In DSDV, energy-goodput reduces from 245 bits/J to173 bits/J but AODV is affected very little by the geographic restrictions present. The energy performance of DSR is better than DSDV and AODV at lower speeds but DSDV performs better as speed increases. AODV shows the worst performance as shown in Figure 5.34. Energy utilized by bo th DSR and AODV increases as speed
73 increases but energy-goodput va lue of DSDV almost remain s constant after a certain speed. Hence reactive routing protocols are mo re sensitive to the variation in speed than proactive protocols. Figure 5.35. Energy-Goodput Analysis of Va rious Routing Algorithms using Gauss Markov Mobility Model For increasing speeds, the energy perf ormance of DSR and AODV with Gauss Markov mobility model is slig htly worse than its perfor mance with Manhattan Grid model. This may be because reactiv e protocols are more affected by temporal dependence of velocity DSDV on the other hand performs slightly better with Gauss Markov model than with Manhattan Grid model. 5.2.2 Energy-Goodput Analysis using Group Mobility Models The effect of spatial dependence of velo city on the energy performance of DSR, AODV and DSDV is analyzed in this section. From Fi gures 5.39 and 5.40 it can be understood that the energy-goodput of DSR drops from about 340 bits/J to around 250 bits/J as the scenario is transformed fr om single group to multiple groups (i.e) as the degree of spatial dependence of ve locity decreases the energy-goodput also
74 decreases. But the energygoodput of the various multiple group scenarios is almost of the same level. Figure 5.36. Energy-Goodput Analysis of DSR using RPGM Mobility Model for Various Speeds Figure 5.37. Energy-Goodput Analysis of DSR using RPGM Mobility Model for Various Group Scenarios
75 From Figures 5.41 and 5.42, as speed increases from 5 m/s to 80 m/s the energygoodput decreases gradually from 336 bits/J to 75 bits/J. Similar to DSR, as the degree of spatial dependency decreases al so the energy-goodput decreases gradually. In case of DSDV the energy-goodput decr eases minimally when considering DSR and AODV. As speed increases from 5 m/s to 80 m/s the energy-goodput decreases from around 245 bits/J to 206 bits/J. Figure 5.38. Energy-Goodput Analysis of AO DV using RPGM Mobility Model for Various Speeds Again in DSDV as the degree of spatial depe ndence of velocity in RPGM decreases, the energy-goodput decreases, but this drop is small when compared to the drop obtained using AODV or DSR. Thus we c onclude that though DSR and AODV have high energy-goodput at lower speeds, DS DV maintains a moderate energy-goodput for all speeds.
76 Figure 5.39. Energy-Goodput Analysis of AO DV using RPGM Mobility Model for Various Group Scenarios Figure 5.40. Energy-Goodput Analysis of DSDV using RPGM Mobility Model for Various Speeds
77 Figure 5.41. Energy-Goodput Analysis of DSDV using RPGM Mobility Model for Various Group Scenarios Figure 5.42. Energy-Goodput Analysis of DS R using Column Mobility Model for Various Speeds
78 Figure 5.43. Energy-Goodput Analysis of DS R using Column Mobility Model for Various Group Scenarios The energy-goodput performance of Column mobility model is similar to those displayed by RPGM. This can be easil y comprehended from Figures 5.45, 5.46, 5.47, 5.48, 5.49 and 5.50. Figure 5.44. Energy-Goodput Analysis of AODV using Column Mobility Model for Various Speeds
79 Figure 5.45. Energy-Goodput Analysis of AODV using Column Mobility Model for Various Group Scenarios Figure 5.46. Energy-Goodput Analysis of DS DV using Column Mobility Model for Various Speeds
80 Figure 5.47. Energy-Goodput Analysis of DS DV using Column Mobility Model for Various Group Scenarios Figure 5.48. Energy-Goodput Analysis of Vari ous Routing Algorithms using Pursue Mobility Model
81 DSR shows excellent energy performance at lower speeds with Pursue mobility model. Even though at higher speeds th e energy-goodput decreases slightly, DSR shows better performance compared to AODV. As discussed previously DSR performs better with mobility models that have more homogeneity. It is believed that as speed increases and the pursuer nodes catch up more closely with the pursued node, there is more closer route forma tions with the pursued node and hence homogeneity decreases (or the degree of sp atial dependence of velocity decreases) and hence the energy performance of DSR di minishes as shown in Figure 5.51. The same explanation applies to the drop in energy-goodput in AODV. DSDV shows the worst performance of the th ree routing protocols. Bu t the energy-goodput of DSDV remains constant for all speeds from 5 m/s to 80 m/s. 5.3 Summary In this chapter we analyze the scalability of MANETs using en tity and group mobility models. We also study the energy performan ce of ad hoc networks in detail. The effect of realistic mobility characteristics on the overall performance of mobile ad hoc networks was explored.
82 Chapter Six Conclusions and Future Work 6.1 Conclusions The protocol performance such as, scalabil ity and energy utiliza tion of a mobile ad hoc network is affected by the movement pattern of mobile nodes in realistic environments. The effect of realistic mobility characteristics such as temporal dependence of velocity, spatial dependence of velocity and geographic restrictions on the protocol performance is studied in de tail. From the performance analysis carried out, the following conclusions can be made: When there are geographic restrictions associated with the movement of a mobile node in a MANET, as in Manhattan Gr id model, or when there is a high degree of temporal dependence of velocity as in Gauss Markov model, there are more link breakages and hence there are more packet s being dropped. When more packets are dropped more retransmissions takes place lead ing to the generation of more control packets and hence lower PDR values as speed increases. One main reason for this performance drop in DSR can be attributed to the fact that control overhead increases more drastically as speed increases from 5 m/s to 20 m/s, when MG model or GM model was used instead of RW model. AODV and DSDV are more stable when operating with mob ility models that have less homogeneity such as MG, GM and also group mobility models with multiple groups. We observe that DSR performs better with mobility models wh ere the nodes have more homogeneity (higher degree of sp atial dependence of velocity) In RPGM model and Column model as the network transforms fr om being a single group scenario to multiple group scenarios, there is lesser homogeneity and hence
83 there is less route formations between nodes in different groups. This results in longer route formations to transmit data across node s of different groups. In Pursue model, the distribution of nodes is more homoge neous and hence there are shorter route formations between nodes. Although DSR has a high PDR and low control overhead at lower speeds, it has to be noted that there is a drastic drop in perf ormance as speed increases and hence we need to consider the consistency of the pr otocol performance when the mobile ad hoc network is scaled to include more numb er of nodes and operate in high mobility conditions. As speed increases more route ch anges take place and hence more routing packets are transmitted which in turn has an effect on the packet delivery of the network. DSR performs better with group m obility scenarios which have high degree of spatial dependence than with those wh ich are less homogeneous (multiple groups). DSDV seems to be more reliable routing protocol to be used in group mobility scenarios with less homogeneity (multiple groups) as there is no severe effect on protocol performance when speed increases or when the degree of spatial dependency decreases. As speed increases, more route changes take place and hence there is a need for more frequent updates of the rout ing table which increas es routing overhead and hence reduces PDR values but in DS DV, when degree of spatial dependency (homogeneity) decreases, all the route forma tions within a group are usually stable even when mobility increases as the nodes are more closely packed among themselves and hence these routes need not be updated frequently. This helps to maintain the control overhead constant or at the least there is less variation of control overhead. Hence even when the single group transforms to multiple group scenarios, there is very little performance drop. AOD V performs similar to DSDV when the degree of spatial dependence of velocity ( homogeneity) decreases, but there is a more visible drop in protocol performance than DSDV.
84 6.2 Future Work Previously, most of the simulations of MANETs were done using Random Waypoint mobility model as a default model. But the scenarios in which Ad hoc networks are implemented are not random in nature as in most cases the m obile nodes are operated by humans whose movements may more likel y follow a certain deterministic pattern. Hence it is necessary to evaluate MAN ET performance with realistic mobility models. Simulations are a valuable tool fo r learning and comparing wireless protocols and techniques, but simulations generally su cceed because we will always be able to find the right protocols and configure it to work well in any particular scenario. Real-world ad hoc networks face problems that don't generally occur in simulation. It is true that unlike simulator experiments, test-bed experiments cannot be perfectly reproduced. Interference and radio propa gation conditions change between each experiment, and are out of th e experimenter's control. Ho wever, experimental results are generally repeatable, and executing th e same experiment many times produces more consistent results . The realistic movement patterns used in our simulations can be integrated into a suitable testbed such as Ad hoc Protocol Evaluation (APE) Testbed  or Network Emula tion Testbed (Netbed) . We can incorporate error models in our simu lations to understand th e effect of packet loss on the performance of the network. By doing this we will be able to mimic the realistic ad hoc environment more effectively. Error model simulates link-level errors or loss by either marking the packet's error flag or dumping the packet to a drop target. We also need to study the effect of node density on the scal ability of a network in any mobility conditions.
85 References  The Wi-Fi Alliance Website, http://www.wi-fi.org  The Wireless LAN Association Website, http://www.wlana.org.  Prasad, N. R.; IEEE 802.11 system design, IEEE Internati onal Conference on Personal Wireless Communications, pp: 490-494, 2000.  Ware, C.; Judge, J.; Chicharo, J.; Dutkie wicz, E.; Unfairness a nd capture behavior in 802.11 ad hoc networks, IEEE Internat ional Conference on Communications (ICC 2000), pp: 159 -163 vol.1, 2000.  Bensaou, B.; Yu Wang; Chi Chung Ko; Fair medium access in 802.11 based wireless ad-hoc networks, First A nnual Workshop on Mobile and Ad Hoc Networking and Computi ng (MobiHOC), pp: 99 106, 2000.  Milanovic, N.; Malek, M.; Davidson, A.; M ilutinovic, V.; Routing and Security in Mobile Ad Hoc Network, IEEE Co mputer Society, pp: 69 73, 2004.  Zhou, L.; Haas, Z.J.; Securing ad hoc networks, IEEE Network, vol-13, pp: 24 30, 1999.  Costa-Requena, J.; Ad-Hoc routing taxonomy and Resource Discovery, Ad Hoc Mobile Wireless Networks Research seminar on Telecommunications Software, 2002.  Muqattash A.; Krunz, M.; CDMA-based MAC protocol for wireless ad hoc networks, Proceedings of the 4th ACM In ternational Symposium on Mobile ad hoc networking and computing (MobiHoc), pp: 153 164, 2003.  Capacity of Mobile Ad-hoc Networks; http://degas.eecs.berkeley.edu/~dtse/adhoc.html
86  Borgonovo, F.; Capone, A.; Cesana, M. ; Fratta, L.; ADHOC MAC: New MAC Architecture for Ad Hoc Networks Provi ding Efficient and Reliable Point-toPoint and Broadcast Services, ACM Wireless Networks, vol:10, pp:359-366, 2004.  Ad Hoc Networks; http:// www.wu.ece.ufl.edu/wireless/ad_hoc_networks.html  Stallings, W.; Wireless Communicatio ns and Networks; First Edition, Prentice Hall Inc., 2003.  Bai, H.; Atiquzzaman, M.; Hone ywell Aerospace; Error Modeling Schemes for Fading Channels in Wireless Communicat ions: A Survey, IEEE Communications, vol. 5, 2003.  Grossglauser, M.; Tse, D. N. C.; M obility Increases the Capacity of Ad hoc Wireless Networks; IEEE/ACM Trans actions on Networking, vol. 10, 2002.  Mobile Ad Hoc Networks (MAN ETs) are not a Fundamentally Flawed Architecture; http://www.dailywireless.org.  Muqattash, A.; Krunz, M.; Power Cont rolled Dual Channel (PCDC) Medium Access Protocol for Wireless Ad Ho c Networks, Proceedings of IEEE INFOCOM, 2003.  Liu, M.; Liu, M.T.; A Power-saving Scheduling for IEEE 802.11 Mobile Ad Hoc Network, Proc. Internationa l Conference on Computer Networks and Mobile Computing (ICCNMC), 2003.  Sarela, M.; Measuring the effects of m obility on reactive ad hoc routing protocols, Helsinki University of Technology Laborat ory for Theoretical Computer Science Research Reports 91, 2004.  To, K. A.; Chawathe, R.A.; Srivastava, R; Functional Analysis of Mobile Ad Hoc Routing Protocol Scalability, www.owlnet.rice.edu/~takhoa/research  The OPNET website; http://www.opnet.com.  The Network Simulator: ns-2 official website; http://www.isi.edu/nsnam/ns  Bansal, S.; Shorey, R.; Misra, A.; Co mparing the routing energy overheads of adhoc routing protocols, Proc. IEEE Wi reless Communications and Networking (WCNC), vol: 2, pp: 1155 1161, 2003.
87  Zheng, Q.; Hong, X.; Ray, S.; Recent a dvances in mobility modeling for mobile ad hoc network research, Proc. 42nd ACM Southeast Regional Conference (ACMSE), pp: 70-75, Apr. 2004.  Hong, X.; Gerla, M.; Pei, G.; Ch iang, C; A group mobility model for ad hoc wireless networks, Proc. 2nd ACM MSWim, pp. 53-60, Aug. 1999.  Chen, B.; Chang, C.H.; Mobility Im pact on Energy Conservation of Ad Hoc Routing Protocols, Project funded by U. S. National Science Foundation Grant # 0227879.  Bettstetter, C.; Resta, G.; Santi, P.; The Node Distribution of the Random Waypoint Mobility Model for Wireless Ad Hoc Networks, IEEE Transactions on Mobile Computing, vol.2, pp: 257-269, 2003.  Takai, M.; Bagrodia, R.; Lee, A.; Gerla, M.; Impact of Channel Models on Simulation of Large Scale Wireless Ne tworks, Proc.7th ACM international symposium on Modeling, analysis and simula tion of wireless and mobile systems, pp: 62 69, 2004.  Sklar, B.; Rayleigh Fading Channels in Mobile Digital Communication Systems Part I: Characterization, IEEE Comm unications Magazine, pp: 90-100, Jul 1997.  Punnoose, R.J.; Nikitin, P.V.; Stancil, D.D.; Efficient simulation of Ricean fading within a packet simulator, 52nd IEEE VTC-Fall, pp.764.767, 2000.  Takai, M.; Martin, J.; Bagrodia, R.; Effects of wire less physical layer modeling in mobile ad hoc networks, Proceedings of the 2nd ACM International symposium on Mobile ad hoc networki ng & computing, pp:87-94, 2001.  Jiang, H.; Garcia-Luna-Aceves, J.J.; Performance Comparison of Three Routing Protocols for Ad Hoc Networks, Proc .10th IEEE International Conference on Computer Communications and Ne tworks (IEEE IC3N), Oct 2001.  Cano, J.C.; Manzoni, P.; A Performan ce Comparison of Energy Consumption for Mobile Ad Hoc Network Routing Protoc ols, Proc. 8th IEEE-ACM International Symposium on Modeling, Analysis an d Simulation of Computer and Telecommunication Systems (MACSOTS), pp: 57, 2001.
88  Jardosh, A.; Belding-Royer, E. M., Al meroth, K.C., Suri, S.; Towards Realistic Mobility Models for Mobile Ad hoc Ne tworks, Proc. MobiCom, pp. 217 -229, Sep. 2003.  Chen, B.; Chang, C.H.; Mobility Impact on Energy Conservation of Ad Hoc Routing Protocols SSGRR 2003, Jul. 2003.  Bai, F.; Sadagopan, N.; Helmy, A.; IMPORTANT: A framework to systematically analyze the Impact of Mobility on Performance of RouTing protocols for Adhoc NeTworks Proc. of IEEE INFOCOM, 2003.  Dricot, J.M.; De Doncker; Zimanyi, D. ; Grenez, Fr; Impact of the physical layer on the performance of indoor wireless networks; Proc. of the Int. Conf. on Software, Telecommunications and Computer Networks, 2003.  Sicher, A.; HiperLAN/2 and the Evolution of Wireless LANs, Emerging Technologies Symposium: Broadband, Wirele ss Internet Access, IEEE, pp: 1, 2000.  IEEE 802.11; http://www.mpirical.com /companion/mpirical_companion.html.  Xu, K.; Gerla, M.; Bae, S.; Effec tiveness of RTS/CTS Handshake in IEEE 802.11 based Ad Hoc Networks, Ad Hoc Networ k Journal, Vol. 1 No. 1, Elsevier Science.  Yu, W.; Lee, J.; DSR-based Energy-awa re Routing Protocols in ad hoc networks; Proceedings of the International Confer ence on Wireless Networks (ICWN), Jun. 2002.  Toh, C.K.; Ad Hoc Mobile Wireless Ne tworks: Protocols and Systems; PrenticeHall Inc, 2002.  Cano, J.C.; Manzoni, P.; Evaluati ng the energy consumption reduction in a MANET by dynamically switching-off network interfaces, 6th IEEE Symposium on Computers and Communications (ISCC'01) 2001.  Li, H.; Yu, D.; Comparison of Ad hoc and Centralized Multihop routing, Wireless Personal Multimedia Co mmunications (WPMC 2002), 2002.
89  Bai, F; Helmy, A.; A Survey of Mob ility Modeling and Analysis in Wireless Adhoc Networks, Book Chapter in the upcoming book .Wireless Ad hoc and Sensor Networks to be published by Kluwer Academic Publishers, 2004.  The BonnMotion Mobility Generator: www.cs.unibonn.de/IV/BonnMotion  Radio Propagation Models; http://www.isi.edu/~wei ye/pub/propagation_ns.pdf  Ghosh, J.; Philip, S.J.; Qiao, C.; O RBIT Mobility Framework and Orbit Based Routing (OBR) Protocol for MANET, Un iversity of Buffalo, CSE technical report.  The ns-2 Manual; http://www.isi.edu/nsnam/ns/doc/ns_doc.pdf  Andrs Lagar Cavilla; Masters Th esis: Robustness of simplified simulation models for indoor MANET evaluation, 2005.  User Manual for IMPORTANT Mobility Tool Generators in ns-2 Simulator; http://nile.usc.edu/important/mobility-user-manual.pdf.  ns-2 Tutorial: http://www.isi.edu/nsnam/ ns/tutorial/nsscript7.html  The BonnMotion Mobility Generator; http://www.cs.unibonn.de/IV/BonnMotion  The BonnMotion Mob ility Generator Manual; http://web.informatik.unibonn.de/IV/Mitarbeiter/dewaal/BonnMotion/README.  The Scenario Generator, http://www.comp.nus.edu.sg/~liqm/scengen  Camp, T.; Boleng, J.; Davies, V.; A Survey of Mobility Models for Ad Hoc Network Research, Wireless Communi cation & Mobile Computing (WCMC) : Special issue on Mobile Ad Hoc Networki ng: Research, Trends and Applications, vol. 2, no. 5, pp. 483-502, 2002.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001680962
007 cr mnu|||uuuuu
008 051205s2005 flu sbm s000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001159
Performance evaluation of mobile ad hoc networks in realistic mobility and fading environments
h [electronic resource] /
by Preetha Prabhakaran.
[Tampa, Fla.] :
b University of South Florida,
Thesis (M.S.E.E.)--University of South Florida, 2005.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 103 pages.
ABSTRACT: Mobile Ad hoc Networks (MANETs) are wireless networks, which consist of a collection of mobile nodes with no fixed infrastructure, where each node acts as a router that participates in forwarding data packets. They are a new paradigm of wireless communications for mobile hosts that are resource-constrained with only limited energy, computing power and memory.Previous studies on MANETs concentrated more on energy conservation in an idealistic environment without taking into consideration, the effects of realistic mobility, interference and fading. The definition of realistic mobility models is one of the most critical and, at the same time, difficult aspects of the simulations of networks designed for real mobile ad hoc environments. The reason for this is that most scenarios for which ad hoc networks are used have features such as dynamicity and extreme uncertainties. Thus use of real life measurements is currently almost impossible and most certainly expensive.Hence the commonly used alternative is to simulate the movement patterns and hence the reproduction of movement traces quite similar to human mobility behavior is extremely important.The synthetic models used for movement pattern generation should reflect the movement of the real mobile devices, which are usually carried by humans, so the movement of such devices is necessarily based on human decisions. Regularity is an important characteristic of human movement patterns. All simulated movement models are suspect because there is no means of accessing to what extent they map reality. However it is not difficult to see that random mobility models such as Random Walk, Random Waypoint (default model used in almost all network simulations), etc., generate movements that are most non-humanlike.Hence we need to focus on more realistic mobility models such as Gauss Markov, Manhattan Grid, Reference Point Group Mobility Model (RPGM), Column, Pursue and other Hybrid mobility models. These models capture certain mobility characteristics that emulate the realistic MANETs movement, such as temporal dependency, spatial dependency and geographic restriction. Also a Rayleigh/Ricean fading channel is introduced to obtain a realistic fading environment.The energy consumed by the data, MAC, ARP and RTR packets using IEEE 802.11 MAC protocol with the various mobility models in fading and non-fading channel conditions are obtained using ns-2 simulations and AWK programs. The realistic movement patterns are generated using three different mobility generators BonnMotion Mobility Generator, Toilers Code and Scengen Mobility Generator.
Adviser: Dr.Ravi Sankar.
Packet delivery ratio.
x Electrical Engineering
t USF Electronic Theses and Dissertations.