Proceedings of 2011 International Conference on Signal Processing, Communication, Computing and Networking Technologies (ICSCCN 2011) A Study on The Effect of Traffic Patterns on Routing protocols in Ad-hoc Network Following RPGM Mobility Model Arindrajit Pal∗ , Jyoti Prakash Singh† , Paramartha Dutta‡ , Paulomi Basu† and Doyel Basu† ∗ Dept. of Computer Science and Engineering, Academy of Technology, West Bengal, India Email: arindrajit@gmail.com † Dept. of Information Technology, Academy of Technology, West Bengal, India, Email: jyotip.singh@gmail.com ‡ Department of Computer and System Sciences, Visva-Bharati University, West Bengal, India, Email: paramartha.dutta@gmail.com Abstract—Mobile Ad-hoc networks(MANET)s are self organizing networks which can form a communication network without any fixed infrastructure. The constant bit rate (CBR) traffic is very well known traffic model for mobile Ad-hoc network. CBR traffic generates data packets at a constant rate. This constant rate packet generational is good enough for text data packets. The multimedia applications generally generate packets which sometimes increase and might be sometimes followed by some idle period. CBR traffic model does not accommodate the specific features of multimedia data. The day by day increase of multimedia data prompts us to investigate some other traffic models which can accommodate the specific features of these data. One such traffic model is Exponential traffic model which is based on exponential distribution. The other model is Pareto traffic model based on Pareto distribution. In this article, we have tried to study the behavior of mobile Ad-hoc network routing protocols under Exponential and Pareto traffic sources for nodes moving with Reference Point Group Mobility model (RPGM). To the best of our knowledge, this is the first attempt to analyze the effect of traffic patterns with Reference Point Group Mobility (RPGM) model on mobile ad hoc network routing. We have chosen the Normalized routing load, Packet Delivery Fraction and throughput as our figures of merit to compare various protocols. We found high Normalized Routing Load for both Exponential and Pareto traffic compared to CBR traffic in both Dynamic Source Routing (DSR) and Ad hoc On-demand Distance Vector protocol. The Packet Delivery Fraction is comparable in all types of traffic patterns and routing protocols. Throughput is high for Ad hoc On-demand Distance Vector (AODV) routing protocol across all traffic models. I. I NTRODUCTION Ad hoc network is a infrastructure-less wireless network. A source node communicates with a destination node directly if the destination is within the transmission range of the source otherwise source uses other intermediate nodes between source and destination as relay nodes. The ad hoc network is a multihop network without any centralized control. Hence, the routing and the resource management are done through different nodes in distributed manner. The physical movement 978-1-61284-653-8/11/$26.00 ©2011 IEEE of nodes leads to breakup of communication links between adjacent nodes which leads to a number of control packet transmission to announce these topology updates. The statistical behaviour of physical movement of nodes are captured by different mobility models like Random Way Point (RWP)[2], [10], [18], [17], Manhattan Grid (MGM)[6], Reference Point Group Mobility Model (RPGM) [7], [19] etc.. Camp et al. [3] have provided a good survey of the most frequently used mobility models. The routing protocols in Ad-hoc network can be grouped into two main categories based on the route maintenance technique. They are (i) Proactive routing protocol and (ii) Reactive routing protocol. A route between each pair of nodes is maintained in proactive routing protocols regardless of whether those routes are really being used or not. In network, where proactive routing algorithm is used, each node has got a complete picture of connectivity of all nodes in the network. To keep this connectivity information, each node maintains a table of information containing the minimum number of hops and the next hop in the direction of a specific destination. This table is kept up to date by broadcasting all topology updates immediately or with a small time shift to all other nodes in the network. So, proactive routing algorithms are also known as table driven algorithm. In proactive routing algorithm every node has a route to destination always available. It does not have to incur any delay for route discovery when a traffic source begins a session with a remote destination. For a given link cost, the nodes maintain and keep an optimal routes to all other nodes of the network. A very popular and early proactive routing protocols is Destination-Sequenced Distance-Vector (DSDV) proposed by Perkins [15] in 1994. Like other table driven routing algorithms it maintains several routing tables which provide information about every possible destination within the network. The tables contain the sequence numbers for every destination along with the minimum number of hops 233 Proceedings of 2011 International Conference on Signal Processing, Communication, Computing and Networking Technologies (ICSCCN 2011) and the next hop in direction of a specific destination. However, due to high overheads and somewhat poor convergence behavior, this protocol is not directly suitable for mobile ad hoc networks. In a network using reactive routing algorithm, nodes find and maintain the route which are really needed. The nodes do not keep the complete knowledge of the network topology. Whenever a path to destination is needed, a route is searched and established. So, such a routing algorithm is also known as on demand routing algorithm. The Ad hoc On-demand Distance Vector (AODV) proposed by Perkins [16] in 1997 is a popular example of on-demand routing protocol. AODV uses traditional routing tables, one entry per destination. Dynamic Source Routing (DSR) [9] is another example of reactive routing which stores the complete hopby-hop route to the destination. The major difference between AODV and DSR is that DSR uses source routing in which a data packet carries the complete path to be traversed. On the contrary in AODV, the source node and the intermediate nodes store the next-hop information corresponding to each flow for data packet transmission. All popular routing protocols for MANET considers the constant bit rate traffic as the traffic source. The traffic patterns play an important role in so far as the performance of a routing protocol is concerned. The traffic pattern changes with the nature of applications. Traditional data applications generate constant bit rate traffic which is the traffic model of choice of most of the researchers for a long time in the area of mobile ad hoc network. Recently multimedia applications have drawn the attention of a lot of researchers in mobile ad hoc network. These multimedia applications have a radically different traffic patterns. The data rate in voice application increases till it reaches a maximum peak. It is followed over an ideal period. This pattern of traffic can be captured by Exponential or Pareto traffic. The rest of this paper is organized as follows. Section 2 includes the Reference point group mobility model(RPGM). Section 3 contains a brief introduction to different traffic patterns. In section 4, we describe two popular reactive routing protocols. Section 5 contains the simulation settings and results. We conclude the article in section 6 with some suggestions regarding future directions. II. R EFERENCE POINT GROUP MOBILITY MODEL (RPGM) The real applications where Reference point group mobility model(RPGM) model can be used properly is the mobility behavior of the soldiers moving together in a group in the battle field. There is a logical center or group leader in each group. The movement of the group leader determines the mobility behavior of all other members in the group. Initially, each member of the group is uniformly distributed in the neighborhood of the group leader. At every time instant, each node has its own speed and movement direction is derived by randomly deviating from that of the group leader. This model realizes the spatial dependency of each node of a group with logical center [12][5]. The movement of the � group leader at time t can be represented by motion ������ . Not only does it define the motion of the group leader 978-1-61284-653-8/11/$26.00 ©2011 IEEE itself, but also it provides the general motion trend of the whole group. Each member of this group deviates from this � by some degree. The motion vector general motion ������ � ������ can be randomly chosen or carefully designed based on certain predefined paths. The movement of group members is significantly affected by the movement of the group leader. For each node, mobility is assigned to a reference point that follows the group movement. Upon this predefined reference point, each mobile node could be randomly placed in the neighbourhood. Formally, the motion vector of group member i at time t, ��� , can be described as, � + � × ��� ��� = ������ (1) where the motion vector �.��� is a random vector deviated by group member � from its own reference point. III. T RAFFIC MODELS IN MANET A. Constant Bit Rate (CBR) The most popular traffic source in network simulation is Constant Bit Rate (CBR) traffic. In case of CBR traffic source, the data packets are generated at a constant rate with parameters to change the inter packet interval. The CBR class is commonly used for normal data transmission. In this traffic, the data rate and the delay remain constant during the packet transmission. The CBR traffic is not very useful for simulation of real time multimedia traffic generated on demand and videoconferencing services [4][1]. B. Exponential Traffic The exponential traffic model follows an ON/OFF packet generation pattern. During On period, the packets are generated following an exponential distribution. One can tune the traffic generation pattern using three parameters specified in this traffic generator. The first parameter denotes the rate at which the traffic can be generated during ON period. The second and third parameter denote the ON and OFF periods respectively. The Exponential Traffic class is found to be quite useful for real time multimedia, video and voice traffic[13], [1]. C. Pareto traffic The Pareto traffic model also follows an ON/OFF packet generation pattern. During ON period, the packet generation rate varies following Pareto distribution. The shape of the Pareto distribution is tuned using its shape parameter. The other parameters of the traffic generators are the rate at which the traffic can be generated during ON period and the ON and OFF periods respectively. The multimedia, video and voice traffic sources can be simulated well using Pareto traffic generator[13], [1] . IV. ROUTING A LGORITHMS IN MANET We have applied our simulation studies on two popular demand routing protocol namely Ad hoc On-Demand Distance Vector (AODV) and Dynamic source routing (DSR). For the sake of completeness the principles of those routing algorithms are given below. 234 Proceedings of 2011 International Conference on Signal Processing, Communication, Computing and Networking Technologies (ICSCCN 2011) A. Ad hoc On-Demand Distance Vector (AODV) Ad hoc On-Demand Distance Vector (AODV) is an effective example of reactive on-demand routing protocol. It uses ondemand approach for finding routes i.e. a route is established only when it is requested by a source node for transmitting data packets to the receivers and these routes are maintained until they are in need by the source. In AODV, each node maintains at most one route per destination. Being a single path protocol, it has to invoke a new route discovery whenever the only path from the source to the destination fails. When topology changes frequently, route discovery needs to be initiated time and again which can be very inefficient. AODV maintains a destination sequence number generated by the receivers and determines an up-to-date path to the destination. A node updates its route information only if the destination sequence number of the current by received packet is greater than the destination sequence number stored at the node. It indicates the freshness of the route accepted by the source. To prevent multiple broadcast of the same packet, AODV uses broadcast identifier number that ensures loop freedom. This is because the intermediate nodes only forward the first copy of the same packet and discard the duplicate copies. When a route to a new destination is needed, the node uses a broadcast route request (RREQ) to find a route to the destination. Nodes that receive the RREQ find out whether they are the destination or whether they have a fresh route to the destination. Then they respond to the RREQ by unicasting a route reply (RREP) back to the source node [8]. A route can be determined when the request reaches either the destination itself, or an intermediate node with a fresh enough route to the destination. Since each node receiving the request keeps track of a route back to the source of the request, the RREP Reply can be unicast back from the destination to the source, or from any intermediate node that is able to satisfy the request back to the source. The figure 1 shows the route establishment of AODV routing protocol. B. Dynamic source routing (DSR) Dynamic source routing (DSR) is another example of reactive routing protocol. It generates the proper route only when packet needs to be forwarded from source to destination. Within the limit of the transmission range, the process of finding a path is only executed when a path is needed by a node. DSR makes aggressive use of source routing and route caching. With source routing, complete path information is available and routing loops can be easily detected and eliminated without requiring any special mechanism. Because route requests and replies are both source routed, the source and destination, in addition to learning routes to each other, can also learn and cache routes to all intermediate nodes. The addresses of intermediate nodes in the route are kept within the delivered packets. The route discovery process broadcasts a ROUTE REQUEST packet that is flooded across the network in a controlled manner. ROUTE REQUEST packets use sequence numbers to prevent duplication. The request is answered by a ROUTE REPLY packet either from the destination node or an intermediate node that has a cached 978-1-61284-653-8/11/$26.00 ©2011 IEEE [h] Fig. 1. Route establishment in AODV route to the destination[6]. To take full advantage of route caching, DSR replies to all requests reaching a destination from a single request cycle. Thus, the source learns many alternate routes to the destination, which will be useful in case the primary route fails. The figure 2 shows the route establishment of DSR protocol. V. C OMPARISON BETWEEN AODV AND DSR Both AODV and DSR are two on demand reactive routing protocols. But these two routing protocols maintain the routes in different way. AODV maintains the destination sequence number and choices to RREQ floods. AODV uses a more rigorous route-error notification mechanism and therefore has access to more accurate routing information. In Figure 1, we have shown that the path is 1-4-6. AODV has a concept of route cache. If some intermediate node holds the path from that node to the destination then AODV may established a path from source node to that route cache. If we consider a route node 2 holds cache route (5-6) then another path is 12-5-6. DSR is source routing protocol. It maintains multiple routes to a destination and uses immoral mode listening. The current specification for DSR does not contain any mechanism for route entry invalidation. This leads to state cache entries, predominantly at high mobility. In Figure 2, we have shown the route establishment in DSR routing protocols. There are three paths from source to destination like 1-4-6, 1-2-5-6, 13-6. Fig. 2. Route establishment in DSR 235 Proceedings of 2011 International Conference on Signal Processing, Communication, Computing and Networking Technologies (ICSCCN 2011) VI. S IMULATION RESULTS AND ANALYSIS Our target in the experiment is to study the behavior of routing protocol with different types of traffic sources under a specific mobility scenario like Reference point group mobility model(RPGM). We use Bonn-Motion [14] for generating RPGM mobility scenarios. We generate four mobility patterns with 30, 40 and 50 nodes moving in an area of 1000�×1000� for a period of 1000 s with the first 3600 sec of each mobility pattern ignored. It has been observed that with the Reference point group mobility model(RPGM), nodes have a higher probability of being near the center of the simulation area, while they are initially uniformly distributed over the simulation area initially. So, we skip 3600 s at the beginning to mitigate the boundary effects of node movement simulation. The cbrgen tool which is a part of ns-2 [11] distribution is used to generate Constant Bit Rate traffic for 1000s with 1 packet/sec per source. The number of sources and destinations were chosen randomly by cbrgen tool. Similar to cbrgen tool we developed our own tool to generate Exponential and Pareto traffic. The Exponential traffic source generates traffic at 2 KB/s during ON period. The average ON and OFF periods are 315 ms and 325 ms respectively. The Pareto traffic source generates traffic at 2 KB/s during ON period. The average ON and OFF periods are 315 ms and 325 ms respectively. The source and destination are chosen randomly in each traffic generator. We have used ns-2 [11] for network simulation and traces are generated in new trace format. To compare the performance of different routing protocols under various traffic, we have used Normalized Routing Load, Packet Delivery Fraction and Throughputs as out metric. Fig. 3. Packet Delivery Fraction (%) in AODV and DSR for different traffic models in RPGM B. Normalized Routing Load The Normalized Routing Load (NRL) denotes the measure of control packets sent and forwarded by router to transmit every byte of data packet. the Normalized Routing Load is given by �� ���� + �� � ��� × 100 (3) ��� = ���� ��� ��� where �� ���� and �� � ��� are the control packets sent and forwarded by the router respectively and ���� ��� ��� is the data packet received by the application. Figure 4 shows the routing overhead in Normalized Routing Load. Normalized A. Packet Delivery Fraction Packet Delivery Fraction (PDF) is an important figure of merit for Ad-hoc network routing protocols. Packet Delivery Fraction is the ratio of the number of data packets successfully delivered to the destinations to those generated by traffic sources. ���� ��� ��� × 100; (2) � �� = ���� ��� ���� where ���� ��� ��� and ���� ��� ���� are the number of data packets received and sent respectively by the application. The Packet Delivery Fraction for CBR, Exponential and Pareto traffic sources with AODV and DSR routing protocol is shown graphically in Figure 3. From Figure 3, we observe that the packet delivery fraction is very high for AODV and DSR routing protocol in CBR traffic. For node 30, the Packet delivery fraction is slightly low due to few number of nodes. In Exponential and Pareto traffic environment in AODV routing protocol, the Packet Delivery Fraction is very high for various number of nodes. The Packet Delivery Fraction in DSR routing for Exponential and Pareto traffic is average. We also observed that if the number of nodes becomes higher then the Packet Delivery Fraction is also high. From Figure 3 it is evident that, the Packet Delivery Fraction increases for large number of mobile nodes. 978-1-61284-653-8/11/$26.00 ©2011 IEEE Fig. 4. Normalized Routing Load(%) in AODV and DSR for different traffic models in RPGM Routing Load is the ratio of the number of control packets propagated by every node in the network and the number of data packets received by the destination nodes. In case of DSR, each intermediate node records the number of packets queued in control packet. The destination node uses this information when selecting the route. Route breaks occur more frequently in DSR because it often uses old routes. Hence, more ROUTE ERROR packets are transmitted, and consequently, more ROUTE REQUESTS are sent to reconstruct routes. This causes the Normalized Routing Load to increase in DSR which is supported by our experimental results as shown in Figure 4 The Normalized Routing Load is maximum for DSR routing in Exponential and Pareto traffic. In case of Constant Bit Rate(CBR) traffic, the packets are generated at 236 Proceedings of 2011 International Conference on Signal Processing, Communication, Computing and Networking Technologies (ICSCCN 2011) a constant rate. So, the Normalized Routing Load is very less for both AODV and DSR routing. C. Throughput of the network Fig. 5. Throughput in AODV and DSR routing for different traffic models in RPGM Within the simulation time, many data packets are forwarded and received by the different nodes. The throughput of the network denotes the average duration of packets per seconds. i.e. the total number of packets flow in the network per second. � ℎ����ℎ��� = ����� ���� ������� ����� ���������� ���� (4) In Figure 5, we have shown that the throughput rate of AODV routing is higher than the DSR in Constant bit rate (CBR),Pareto and Exponential traffic model. The throughput is very low and almost same in DSR for all types of traffic models. This is because the routing overhead is high in DSR than the AODV routing protocol. This has led to the frequent route failures and the protocol needs to search for next alternate route to the destination in DSR rather than in AODV routing protocol. This results in low throughput. VII. C ONCLUSION In this article, we have tried to analyze the behavior of reactive routing protocols for different nodes following Reference Point Group Mobility model. Along with popular Constant Bit Rate traffic, we have also simulated the Exponential and Pareto traffic sources. The exponential and Pareto traffic sources may be congenial for multimedia applications. We found through our simulations that the packet delivery fraction for reactive routing remains same across all traffic patterns. The throughput increases for exponential traffic in AODV routing. The normalized routing loads on nodes have increased with Exponential and Pareto traffic patterns. The current study have considered only the reactive routing protocol, the same study can be done for proactive routing protocols like Destination Sequenced Distance Vector (DSDV) routing algorithm also. The authors are currently engaged in supplementing the current routing protocols for multimedia application using Exponential and Pareto traffic sources. 978-1-61284-653-8/11/$26.00 ©2011 IEEE R EFERENCES [1] A. Al-Maashri and M. Ould-Khaoua, “Performance analysis of manet routing protocols in the presence of self-similar traffic,” in 31st IEEE Conference on Local Computer Networks, Tampa, Florida, USA, November 2006, pp. 811–818. [2] C. Bettstetter, “Topology properties of ad hoc networks with random waypoint mobility,” ACM SIGMOBILE Mobile Computing and Communication Review, vol. 7, no. 3, pp. 50–52, July 2003. [3] T. Camp, J. Boleng, and V. Davies, “A survey of mobility models for ad hoc network research,” Wireless Communications and Mobile Computing (WCMC): Special issue on Mobile Ad Hoc Networking: Research, Trends and Applications, vol. 2, no. 5, pp. 483–502, 2002. [4] M. U. Chowdhury, D. Perera, and T. Pham, “A performance comparison of three wireless multi-hop ad-hoc network routing protocols when streaming mpeg-4 traffic,” in 8th International Multitopic Conference, December 2004, pp. 516–521. [5] S. K. M. Harminder S. Bindra and A. L. Sangal, “Performance evaluation of two reactive routing protocols of manet using group mobility model,” IJCSI International Journal of Computer Science, vol. 7, May 2010. [6] S. M. D. Himabindu Pucha and Y. C. Hu, “The performance impact of traffic patterns on routing protocols in mobile ad hoc networks,” Computer Networks, vol. 51, pp. 3595–3616, 2007. [7] X. Hong, M. Gerla, G. Pei, and C.-C. Chiang, “A Group Mobility Model for Ad Hoc Wireless Networks,” Workshop on Modelling and Simulation of Wireless and Mobile Systems (MSWiM), pp. 53–60, 1999. [8] G. Jayakumar and G. Ganapathi, “Reference point group mobility and random way point models in performance evaluation of manet routing protocols,” Journal of Computer Systems, Networks, and Communications, vol. 2008, December 2008. [9] D. B. Johnson, D. A. Maltz, and J. Broch, “Dsr: The dynamic source routing protocol for multi-hop wireless ad hoc networks,” in In Ad Hoc Networking, edited by Charles E. Perkins, Chapter 5. Addison-Wesley, 2001, pp. 139–172. [10] P. Lassila, E. Hyyti, and H. Koskinen, “Connectivity properties of random waypoint mobility model for ad hoc networks,” in MedHocNet, le de Porquerolles, 2005, pp. 159–168. [11] S. McCanne and S. Floyd, “ns Network Simulator,” http://www.isi.edu/nsnam/ns/. [12] M. R. Mishra, M. R. Panda, and S. K. Bisoyi, “Topographical automation of manet using reactive routing protocols,” in international conference [ICCT-2010], December 2010. [13] D. Nitnaware and A. Verma, “Performance evaluation of energy consumption of reactive protocols under self- similar traffic,” International Journal of Computer Science and Communication, vol. 1, pp. 67–71, January 2010. [14] U. of Bonn, “Bonnmotion - a mobility scenario generation and analysis tool,” http://www.cs.uni-bonn.de/IV/bomonet/BonnMotion.htm. [15] C. E. Perkins and P. Bhagwat, “Highly dynamic destination-sequenced distance-vector routing (dsdv) for mobile computers,” ACM SIGCOMM Computer Communication Review, vol. 24, pp. 234–244, 1994. [16] C. E. Perkins and E. M. Royer, “Ad-hoc on-demand distance vector routing,” in 2ND IEEE WORKSHOP ON MOBILE COMPUTING SYSTEMS AND APPLICATIONS, 1997, pp. 90–100. [17] J. P. Singh and P. Dutta, “Temporal behavior analysis of mobile ad hoc network with different mobility patterns,” in ICAC3 09: International Conference on Advances in Computing, Communication and Control. ACM, 2009, pp. 696–702. [18] J. P. Singh and P. Dutta, “Temporal modeling of node mobility in mobile ad hoc network,” Journal of Computing and Information Technology, vol. 18, no. 1, pp. 19–29, 2010. [19] J. P. Singh and P. Dutta, “Temporal modeling of link characteristics in mobile ad hoc network,” Journal of Computing and Information Technology, 2011, accepted. 237