Analysis of Network Traffic in Ad-Hoc Networks based on DSDV Protocol

Analysis of Network Traffic in Ad-Hoc Networks based on DSDV Protocol with Emphasis on Mobility and Communication Patterns Vahid Garousi Department of Systems and Computer Engineering Carleton University, Ottawa, Canada vahid@sce. carleton. ca Abstract. An ad-hoc network is a collection of mobile nodes dynamically forming a temporary network without the use of any existing network infrastructure or centralized administration. Because of the limited communication range among mobile nodes in ad-hoc networks, several network hops may be needed to deliver a packet from one node to another one in the wireless network. In recent years, a variety of different routing protocols addressing multi-hop ad-hoc networks have been presented, and their performance issues are discussed and some comparisons are made among them. However, no works have been reported on such analysis with emphasis on mobility and communication patterns of the mobile nodes. W\e are aiming to do such an analysis through simulations in this work. Our simulations are based on ad-hoc networks using DSDV protocol and are done using NS N\etwork Simulator) tool. We give a background on DSDV and present our simulation environment. The analysis is then compiled using the simulation results. Keywords. ad-hoc networks, DSDV, network traffic, mobility, communication patterns. INTRODUCTION An ad-hoc network is a collection of mobile nodes dynamically forming a temporary network without the use of any existing network infrastructure or centralized administration. Because of the limited communication range among mobile nodes in ad-hoc networks, several network hops may be needed to deliver a packet from one node to another one in the wireless network. In recent years, a variety of different routing protocols addressing multi-hop ad-hoc networks have been proposed, such as DSDV (Destination Sequenced Distance Vector) [], TORA (Temporally-Ordered Routing Algorithm) [], AODV (Adhoc On-demand Distance Vector) [3] and DSR (Dynamic Source Routing) [, 5]. Furthermore, the performance issues of the existing routing protocols are discussed and compared in various works, for instance [6]. However, to the best knowledge of the author, no works have been reported on the analysis of network traffic when using DSDV with an emphasis on the mobility of the nodes and the communication patterns among them. What we mean by the mobility of the nodes is to analyze the DSDV protocol's performance such as routing overhead and rate of lost packets when nodes are moving with different speeds and have different pause times between two consecutive moves. We further take into account the scene size of the nodes mobility area, This article is a short version of a report for a graduate course project when the author was with the University of Waterloo, Canada. The full version of the article can be downloaded from: * 0-703-9 79-9/05/$0.00 005 IEEE. number of the nodes, number of connections and also the transmission rate among nodes. Our analysis in this work is based on simulations, for which we use the Network Simulator (NS) package [7]. NS is a state-of-the-art package from the University of Southern California's Information Sciences Institute. The NS package has been used in various works in the literature and that has been the reason why we use it for our simulations. The goal of our simulations is to measure the ability of DSDV routing protocol to react to multi-hop ad-hoc network topology changes in terms of scene size, mobile nodes movement, number of connections among nodes, and also the amount of data each mobile node transmits. To measure this, our basic methodology is to define a set of movement scenarios and communication patterns and apply them to an ad-hoc network. In fact, testing with each data packet originated by a sender mobile node, whether the DSDV routing protocol is able to route and deliver it to the destination node. The rest of this article is structured as follows. In Section, we give a brief introduction on the DSDV protocol and the mobile networking in NS. Section 3 provides the details of our simulation model and environment. Simulation results are then presented in Section. Conclusions are finally drawn in Section 5. BACKGROUND This section presents a brief background information on the DSDV routing algorithm [] and mobile networking in the NS package [7].. DSDV DSDV is a hop-by-hop vector routing protocol requiring each node to periodically broadcast routing updates. One key advantage of DSDV over traditional vector protocols is that it guarantees loop-freedom. Each DSDV node maintains a routing table for the "next hop" to reach a destination node. DSDV tags each route with a sequence number and considers a route R more favorable than R' if R has a greater sequence number than, or if the two routes R and R' have equal sequence numbers but R has a lower metric (such as transmission cost). Each node in the network advertises a monotonically increasing even sequence number for itself. When a node B decides that its route to a destination D has broken, it advertises the route to D with an infinite metric and a sequence number one greater than its sequence number for the route that has broken (making an odd sequence number). This causes any node A routing packets through B to incorporate the

infinite-metric route into its routing table until node A hears a route to D with a larger sequence number [].. Mobile Networking in NS NS (Network Simulator) [7] is a discrete event simulator targeted at networking research. NS is written in C++, with an OTcl/Tk interpreter as a front-end. NS provides substantial support for simulation of TCP, routing, and multicast protocols over wired and wireless (local and satellite) networks. NS supports mobile networking by porting the CMU's Monarch group's mobility extension []. As discussed in Section 6 of NS manual [7], the mobile networking support is realized by a basic wireless model in NS. This mode has components to model mobile nodes and create network topology, movement scenarios, MAC (Modicum Access Control) layer protocols, and different routing protocols. The routing protocols currently supported by NS are DSDV [], TORA [], AODV [3] and DSR [, 5]. For every simulation done with NS, a trace file is generated as output which includes the exhaustive information of all network components ordered by time. Furthermore, NS has a tool for the visualization of the generated trace files, entitled NAM (Network AniMator). NAM is a Tcl/Tk based animation tool for viewing network simulation traces and real world packet trace data. A sample screen -shot of NAM is shown in Figure. I II I vs X Figure -A sample screenshot of NAM animating a simple ad hoc network topology. There are five mobile nodes (labeled 0...). Node is sending a packet to node through a wireless channel. The circles around =- nodes and represent wireless propagation. 3 SIMULATION ENVIRONMENT The goal of our simulations is to measure the ability of DSDV routing protocol to react to multi-hop ad-hoc network topology changes in terms of mobile nodes movement, the number of communication links and also the amount of data each mobile node transmits. We assume fixed physical radio characteristics for mobile nodes in different simulation runs, while changing the network and traffic properties which will be addressed in Section 3.3. We set the physical radio parameters of the network interface for all the mobile nodes in the system to the values shown in Table. In order to perform an ad-hoc network simulation by NS, the nodes movement scenario as well as communication pattern should be specified. Accordingly, we consider two groups of simulation parameters for an ad-hoc network in our analysis: () movement of mobile nodes in an ad-hoc scene and () communication pattern among nodes. In the following, we first present a model for movement scenario and scene size of a network in Section 3.. Section 3. presents the model we use for the communication pattern among nodes in our simulations. The simulation parameters and metrics are given in Sections 3.3 and 3., respectively. We then discuss the simulations execution scenario shortly in Section 3.5. We refer to the output of NS simulations as trace files. In order to get statistical data from the raw trace files, we developed a log -analysis tool, which we discuss in Section 3.5. Parameter Channel Type Propagation Model Network Interface Model MAC Network Interface Queue Type Antenna Model Value Wireless Channel Two Ray Ground Wireless IEEE 0. Drop Tail Omni Directional Table - Network interface parameters of the mobile nodes in our simulations 3. Movement Scenario All of our simulations are done using user-defined nodes movement scenario files. To better distinguish the scenario files, we use the following naming scheme for the movement scenario files: scen-lengthxwidth-nodes-pausetime-maxspeed where Length and Width are the size of simulation scene, which is a rectangular flat area (we assume - dimentional spaces for our simulations), in meters, where mobile nodes are allowed to roam around, Nodes is the number of mobile nodes in simulation, PauseTime is the pause time of mobile nodes between two consecutive movements in seconds, and MaxSpeed is the maximum speed for mobile nodes movement. For example, the file name scen-67ox670-50 -00-0 indicates a scenario file for a scene of size 670x670 m, where 50 nodes are roaming around with maximum speed of 0 m/s and the pause time between consecutive moves of a node is 00 seconds. Movement scenario files in NS are created using setdest utility, in which the above parameters in addition to a total simulation time are given as inputs. 3. Communication Pattern As the basic need of an ad-hoc network in to enable wireless communication among mobile nodes, so the model that defines this wireless communication should be defined in order to simulate an ad-hoc network. There are different network communication models in the base NS, such as CBR (Constant Bit Rate) and VBR (Variable Bit Rate) models. But to the knowledge of the author, there is currently only one model available to wireless and ad-hoc networking support in NS. That is the CBR model. In this model, mobile nodes transmit data packets to each other in a constant bit rate. This model simplifies the simulation; but, in practice, there are other complicated

models like VBR model, in which nodes are free to change their packet transmission rate. CBR mobile communication patterns in NS are generated using cbrgen.tcl utility (located in ns-homedirectory/indep-utils/cmu-scen-gen/). To better distinguish the communication pattern files, we use the following naming scheme for them: cbr-nodes-seed-maxconn-transmissionrate where Nodes is the number of mobile nodes in simulation, Seed is the random number generating seed and MaxConn is the maximum number of connections taking place in the time of the simulation and TransmissionRate is the rate of packet transmission, i.e. number of packets to be transmitted from source mobile nodes in each second. For example, the filename cbr-50--- stands for the communication pattern file for a CBR communication pattern among 50 nodes with maximum connections transmission rate of packets per second. 3.3 Simulation Parameters We consider two groups of simulation parameters for an ad-hoc network in our simulations: () network scene and the movement of mobile nodes in an ad-hoc scene and () the communication pattern among nodes. The following parameters are considered for the ad-hoc scene and nodes movement scenarios. * Scene area size: The flat area size, in length and width, in which mobile nodes can move. * Number of mobile nodes: Number of mobile nodes taking part in the simulation and roaming around in the scene area. * Pause time: Time, in seconds, for which mobile nodes pause between their consecutive movements. The less this time is, the more active the mobile nodes are in moving. * Max speed: Maximum speed of mobile nodes. Mobile nodes are allowed to roam around in the simulation scene area with a O<speed<MaxSpeed. For the communication pattern, we consider the following parameters. * Connections: The maximum number of simultaneous connections taking place in the time of the simulation. * Packet transmission rate: The rate of packet transmission from each source node to a destination node. For simplicity, the packet sizes are assumed to be 5 bytes in all simulations. The simulation results in Section will be grouped by the above simulation parameters. 3. Simulation Metrics While NS provides extensive simulation metrics, we choose the following three metrics as indicators of the DSDV's performance in our simulations. Other metrics can be easily extracted from the simulation output. * Number of total transmitted packets: The total number of packets with different types: Sent, Received, Forwarded and Dropped, which were transmitted between mobile nodes in the period of simulation. This metric provides us with an overview of how the simulated ad-hoc network, with the defined parameters, react to topology changes while nodes are moving. * Packet delivery ratio: Ratio of the number of received packets over sent ones. This metric actually tells us how much reliable our ad-hoc network is. The greater this ratio is, the more reliable the ad-hoc network will be. We investigate the behavior of this metric by changing different operating simulation parameters of the ad-hoc network under study. * Routing Overhead (Ratio of forwardedlsent packets): Routing overhead is the ratio of forwarded (hopped) packets to sent ones. Hopping always has a cost in adhoc networks, and the goal is to minimize it as much as possible. In Section, we will analyze the above simulation metrics for each of our simulation scenarios. 3.5 Execution of Simulations and Trace Analysis The Tcl module dsdv-simulation.tcl, part of the NS package, was used as the process running the simulations in this work. We used a Linux RedHat box with 733 MHz CPU speed and 5 MB of RAM. For each simulation scenario, the movement scenario and communication pattern files were given as inputs to NS running the dsdv-simulation.tcl module. Because the output trace files, created by simulations using module dsdv-simulation.tcl in NS, were raw output log files and did not have enough statistical data for our analysis, we developed a C program to read the raw data from the trace files and generate the simulation metrics, mentioned in Section 3.. Our trace analyzer functions like a parser, in the sense that it reads each line of the trace file, and looks for keywords of packet types (e.g. 's' for sent, 'f' for forwarded, etc.). It then uses these data to generate a set of statistical information. Part of those statistical data will be shown in Section. Details of our NS trace analyzer are presented in [9]. SIMULATION RESULTS As discussed in Sections 3. and 3., we conduct different simulations by changing the parameters for mobile nodes movement scenarios and their connection patterns. We statistically analyze the output of the simulations in terms of the metrics using a trace analyzer we designed and developed specifically for this purpose. In this section, we present the simulation results grouped by the simulation parameters, as listed in Section 3.3.. Scene Area Size As can be seen in Figure -(a), the number of forwarded packets increases as the size of the ad-hoc network scene

area increases. This is intuitive since packets have to be forwarded more frequently in a bigger area to be delivered to typically "far" destination nodes. The same reasoning holds for the packet overhead in Figure -(b). Some packets get lost in this case when the area size is loooxloo im. 000 D 000 I-Sent -0 Received 500- -Dropped a _-Forwarded 0 xri looxlo0 50x50 500x500 670x670 00x00 loooxlooo Square area size (widthxheight in mxm) _- CY E _- 0. - Packet deliveryrate 0. - ; Routing overhead 0 looxlo0 50x50 500x500 670x670 00x00 loooxlooo Square area size (widthxheight) (b)-packet delivery rate and routing overhead Figure - Simulation results for different scene area sizes (Movement scenario files: scen-xxy-50-500-0, x and y variable with connection pattern file: cbr-50---).. Number of Mobile Nodes Fewer packets need to be forwarded when there is larger number of mobile nodes in a scene (Figure 3). Conversely, ratio of lost packets decreases with an increase in numbers of mobile nodes. 000 * 500 - X-Sent -0- Received -Droppe 000 f rwoare 500-00 00 00 600 00 Number of mobile nodes Figure 3 -Number of transmitted packets as a function of number of mobile nodes (Movement scenario files: scen-670x670-n-500-0, n variable with connection pattern files: cbr-n- - -)..3 Pause Time With eight simultaneous connections in the network, Figure -(a) indicates that when nodes make a longer pause between two consecutive moves, ratio of forwarded packets increases. However, when the number of connections in the network is 0, packet overhead (ratio of forwarded packets) does not change considerably with changes in nodes' pause time, Figure - (b) and (c). We can say that when fewer number of simultaneous connections are allowed and mobile nodes tend to be in a fixed position for a longer time (increased pause time), this makes the nodes more active in playing the act of a forwarding mobile node, a node that hops received packets to the next hops (or the destinations) to help them reach their destinations. In the opposite side of this, when nodes are in fixed positions for a little time (decreased pause time), they may frequently exit the direct communication range of the neighboring nodes and this makes them weak in playing the role of a forwarding mobile node, and hence the lower total number of forwarded packets will be yielded. 500-0.6-000- 500 000 -A--Sent 0- Received : 500- - Dropped o -Forwarded Pause Time (seconds) - connections (a) -Number of transmitted packets (number of connections=). 0000I 000-6000- 000 - -A-Sent 000 - --Received Dropped A-Forw arded 0- Pause Time (seconds) - 0 connections (b)-number of transmitted packets (number of connections=0). - 0. - 0.6- A A 0.- 9PD rate ( con) -l PF rate (con) 0.- PD rate (0 con) I PF rate (0 con) Pause Time (seconds) (c)-packet delivery ratio and routing owrhead (number of connections= and 0). Figure -Simulation results for different pause times ofmobile nodes (Movement scenario files: scen-670x670-50-p-0, p variable with connection pattern file: cbr-50---). 000-500 - t000 Sent - --Received 500 -f -Dropped -I -Forwarded 5 0 5 0 Max Speed (meters/second) Figure 5-Number of transmitted packets as a function of maximum movement speeds of nodes (Movement scenario files: scen-670x670-50- 500 -s, s variable with connection pattern file: cbr-50- --).

. Maximum Speed Simulation results shown in Figure 5 point out that when nodes are moving faster, fewer packets are dropped (lost) and the routing overhead is less than the case when the node are moving with less speed..5 Number of Connections Increasing the number of connections among fixed number of nodes enhances the routing overhead and the packet delivery rate, as depicted by Figure 6-(b). 00 00- -n-sent 5 000-0 ~~~Received ;600000 00600 00 00 - --&-Dropped --&-Forw arded.i We presented an aialysis of network traffic in ad-hoc networks based on the DSDV protocol with an emphasis on mobility and communication patterns of the nodes. Our simulations measured the ability of DSDV routing protocol to react to multi-hop ad-hoc network topology changes in terms of scene size, mobile nodes movement, number of connections among nodes, and also the amount of data each mobile node transmits. [] 0.0.95-5 CONCLUSIONS REFERENCES of connections among mobile nodes.6 Transmission Rate Increasing the transmission rate in an ad-hoc network with fixed size and number of node increase the number of transmitted packets in different groups (Sent, Received, Dropped and Forwarded), Figure 7- (a). However as Figure 7-(b) shows, interestingly, this increase does not affect the packet delivery rate nor the routing overhead.,l [] 0.5 Packet delivery rate - 0. [3] of connections among mobile nodes (b)-packet delivery ratio and routing overhead. Figure 6- Simulation results for different number of connections between mobile nodes (Movement scenario file: scen-670x670-50-6000, with connection pattern files: cbr-50 - -c-, c variable). 3500 a [] - [5] 3000-5 500000500-, 0~ 000 500 v [6] I _ 0.950.95 3 0.5 - Packet delivery rate 0.Routing overhead 0 [7] 3 Transmission Rate (packets/second) (b)- Packet delivery ratio and routing overhead. 7-Simulation results for different transmission rates between Figure nodes (Movement scenario file: scen-670x670-50-600-0, with connection pattern files: cbr-50---r, r variable). [] [9] C. E. Perkins and P. Bhagwat, "Highly dynamic Destination-Sequenced Distance-Vector routing (DSDV) for mobile computers," presented at Proceedings of the SIG-COMM '9 Conference on Communications Architectures, Protocols and Applications, 99. V. D. Park and M. S. Corson, "A highly adaptive distributed routing algorithm for mobile wireless networks," presented at In Proceedings of INFOCOM'97, April 997. C. E. Perkins and E. M. Royer, "Ad hoc On-Demand Distance Vector Routing," presented at Proceedings of the nd IEEE Workshop on Mobile Computing Systems and Applications, New Orleans, LA, February 999. D. B. Johnson, "Routing in ad hoc networks of mobile hosts," presented at Proceedings of the IEEE Workshop on Mobile Computing Systems and Applications, December 99. D. B. Johnson and D. A. Maltz, "Dynamic source routing in ad hoc wireless networks," Mobile Computing, edited by Tomasz Imielinski and Hank Korth, Kluwer Academic Publishers, pp. 53-, 996. D. B. Johnson, Y. H. J. Broch, D. A. Maltz, and J. Jetcheva, "A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols," presented at Proceedings of the Forth Annual ACM/IEEEE International Conference on Mobile Computing and Networking (MobiCom'9), 99. Kevin Fall and K. Varadhan, "The ns Manual," University of Southern California, Information Sciences Institute (ISI), dioumentation.htmi 00. CMU Monarch Project, "Monarch Project Extensions to ns-. h accessed November 00. V. Garousi, "Simulating Network traffic in Multi-hop Wireless Ad Hoc Networks based on DSDV protocol using NS Package," Course Project Report for E&CE750 "Software and Protocols in Mobile Systems", Dept. of Elec. and Comp. Eng., University of Waterloo, Fall 00.