Analytical Study of the Effect of Transmission Range on the Performance of Probabilistic Flooding Protocols in MANETs

Muneer Bani Yassein, Qusai Abuein, Mohammed Shatnawi, Deya Alzoubi Faculty of Computer Information Technology, Jordan University of Science and Technology, Jordan
Abstract: Broadcasting is one of the most important operations that are used in Mobile Ad Hoc Networks (MANETs) to disseminate data throughout the entire network. Simple flooding is the conventional operation that performs broadcasting in MANETs. Although flooding is a simple operation that achieves high delivery of data, it has many disadvantages summarized by the redundant broadcasts, contention and collision, which are referred to as the broadcast storm problem. Probabilistic protocols stand to provide a good solution to the problems associated with simple flooding .This paper, presents a comprehensive analytical study for the performance of probability-based routing protocols under different transmission ranges, and it shows the effect of this parameter on the overall performance metrics. All experiments are conducted using NS2. The results show that when the transmission range values and number of Probability P increases the performance of the multiple-Ps algorithms improved, where the protocol with higher P value (P4) outperforms all other protocols in terms of Packet Delivery Ratio (PDF), End-To-End Delay (ETED) and the routing overhead Keywords: MANET, broadcasting, flooding, probabilistic flooding.

1. Introduction
A Mobile Ad Hoc Network (MANET) is regarded to be a collection of mobile nodes that communicate with each other without depending on any pre-existing infrastructures [12]. MANETs are made up of mobile and autonomous nodes that are connected via wireless links and act as both hosts and routers. Hosts are free to move randomly and organize themselves arbitrarily. Thus, the networks topology may change rapidly and unpredictable in a manner. MANETs may operate in a standalone fashion or be connected to the larger Internet [17]. MANETs have many fundamental properties such as: dynamic network topology, multihop routing, distributed operation and autonomous terminal. Mobile Ad hoc networks can be used in numerous situations and can provide tremendous opportunities, particularly whenever there is a need for establishing a network for a limited period of time and where a wired infrastructure maybe nonexistent or very difficult to be deployed. The applications of MANETs include search and rescue, academic, industrial applications and Personal Area Networks (PANs). Broadcasting is one of the most important mechanisms that are used in MANETs, where the source node sends the same packet to all nodes in the network. There were many models designed to achieve this schema such model are one-to-one model, where each transmission is directed to only one neighbour using narrow beam directional antennas or separate frequencies for each node.

Another model is the one-to-all, where transmission occurs by sending a packet from a node to reach all nodes that are within its transmission area [14, 8, 1]. In fact, broadcasting is more closed for the one-to-all model than one- to- one model. The one-to-many is another model, where frozen or changeable angular beam antennas can be used to reach several neighbours at once [14]. Flooding is one of the most deployed mechanisms that are used to perform broadcasting [10]. According to this mechanism, each node transmits the same packet many times for all its neighbours. Therefore, Redundancy, collisions and contention are the most challenging problems that are encountered by flooding. These three problems are collectively referred to as the broadcast storm problem [15]. One of the possible suggested solutions that could be applied to prevent such problems is to memorize the packets that are received during the flooding operation, and to prevent sending repeated copies of the same packet [16].

2. Related Work
In the literature, many approaches have been proposed to perform broadcasting. Williams and Camp [8] have classified broadcasting protocols into: simple flooding, probabilistic-based, counter-based, distancebased, location-based and neighbour-knowledge-based schemes. In simple flooding [9], a nave flooding approach is performed through the entire network, where each

mobile host broadcasts a received packet provided that it has not been broadcasted it before. Packets that have already been received are discarded. In probabilistic-based schemes, upon receiving a broadcast packet for the first time, the node rebroadcasts it again with a probability P. When P =1, the probabilistic based approach converges to simple flooding. Sasson et al [16] and Zhang et al [13] proposed a probabilistic approach to reduce the redundant transmission of packets encountered in the simple flooding and to alleviate the broadcast storm problem. In this approach, a pre-determined broadcast probability P is assigned to each node in advance. Upon receiving a packet, each node rebroadcasts that packet according to the specified probability. Another approach described by Cartigny et al [11], where computes the probability P is computed based on the number of neighbour nodes and a fixed efficiency parameter K. This approach performs well in terms of reachability of broadcast; however, it has a shortcoming in that the computation of the probability P depends on constant data, including constant efficiency parameters and local density. Counter-based schemes [16] prevent the rebroadcast if their packets have already been heard for many times in particular period of time. Distance-based schemes [8] prevent the packet rebroadcasts when distance between the source and destination is less than a given threshold. In location-based scheme, a node rebroadcasts a packet with respect to additional coverage concepts. In selecting forwarding neighbours a broadcasting node elects some of its 1-hop nearest nodes as rebroadcast nodes.

time, if so then the node will discard the RREQ; otherwise the node will find the number of its neighbours. If the number of its neighbour is less than the average then the node is in sparse area. Therefore, high broadcast probability is assigned to the RREQ. On the other hand if the number of neighbours is greater than the average then the node is in dense area. Hence, a low probability is assigned to RREQ. Bani Yassein et al in [2, 5] use the concept of four P's (4P) in their smart probabilistic to calculate the rebroadcasting probability based on local information of the neighbours. Also, they chose the value of P's such that P1>P2>P3>P4 without specifying initial value of P. This technique is performed by calculating the average number of neighbours (avg) in the network and calculates the average number of neighbours for nodes whose number of neighbours is less than average (avg1) and the average number of neighbours for the nodes whose number of neighbours is greater than average (avg2). When receiving a RREQ for the first time, the node gets the values of avg, avg1 and avg2 then it gets the number of its neighbours C. Now if C < avg1, then this node is in low sparse area, which means high probability is assigned to RREQ P= P1. If avg1 C <avg then the node is in medium sparse area and a medium high probability is assigned to RREQ P=2P. If avg < C avg2 then the node is in medium dense area and medium low probability is assigned to RREQ P=P3. Finally if C avg2, then the node is in high dense area and low probability is assigned to RREQ P=P4.

4. Experiments and Results. 3. Methodology

Bani Yassein et al in [4] use the concepts of Fixed Probabilistic (FP) to rebroadcast the packet in which the intermediate node rebroadcast the packet according to a per-determined probability P regardless of the number of neighbouring nodes in literature this probability has been chosen as 0.7. Bani Yassein et al in [3, 7, 6] use the adjusted probabilistic algorithm (2P) to control the rebroadcasting process. In this algorithm, the probability of rebroadcasting depends on whether the node is in dense area or in sparse area. The methodology is to compare the number of neighbours with the average number of neighbours in the networks and then assign a low probability to the intermediate node if it is in dense area and assign high probability if its in sparse area. This technique is performed by calculating the average number of neighbours in the network. The neighbours of node A are defined as the nodes that can be directly reached by node A. Upon receiving a Route Request (RREQ), the node checks, whether the packet is not received for the first

4.1.Experimental Environment
All experiments are implemented and evaluated using the Network Simulator 2 (NS-2) version 2.33 running under Linux operating system. NS-2 is a discrete event simulator targeted at networking research. NS-2 provides substantial support for simulation of TCP, routing, and multicast protocols over wired and wireless (local and satellite) networks.

4.2.Simulation Parameters
Using NS-2, we have carried out a series of experiments to study the performance of the probability-based protocols using different values of probability P. The probability thresholds that we have used are P1=0.7 which we denoted to as FP (Fixed Probability), 2P (Adjusted probabilistic podcasting) where we used two different P values and 4P (smart probabilistic podcasting) where we used four different P values all are based on the number of neighbour node. In addition, we studied the effect of different

transmission ranges that vary from 100, 150, 200, 250, to 300. All of the experiments carried out in a network topology of size 600*600 m, 50 mobility nodes, 4 packet generation rates, 900 seconds simulation time, and 8 m/s maximum speed. Table 1 has shown the different configuration values that were used in all the performed simulations. To ensure the accuracy of our experiments we have run 15 scenarios per each FP, 2P and 4P probabilities with different transition range values.
Table 1: The Simulation Parameters.

figure it is clear that the probabilistic algorithm using 4P achieved the best results for all transmission ranges higher than 150 m. In addition, the higher the transmission ranges, the better the performance of 4P protocol and all other protocols. Also, as the probabilistic approaches use more values of P, the protocol performs better delay. That is, 4P is better than 2P and better than FP. This is due to control flooding and by reduces processing time (i.e. breakage links and queuing) that was used during sent data packet form source to destination.

Parameter Routing Protocol Network dimension Number of node Bandwidth Packet generation rate Packet size Simulation time Max speed Pause time Number of connection Data Traffic Transmitter range

Value AODV 600*600 meter 50 2Mbit 4 512 bytes 900 seconds 8 meters/second 0 sec 10 CBR 100,150,200, 250, 300 metes
Figure 1: Average ETED vs. Transmission Range.

4.3.2.Packet Delivery Fraction (PDF)

Figure 2 shows the effect of different transmission range values on PDF regarding to different probabilistic algorithms FP, 2P, and 4P. When the transmission range values and number of P increase the PDF values are also increase. From the figure, it is shown that beyond a transmission range of 200 m, the packect delivery fraction increases significantly for all protocols because the number of drop packets resulting from unstable route used is reduced.

4.3.Experimental Results
We evaluated the performance of the probability-based routing protocols with multi-probability values using different performance metrics such as the average ETED, Packet Delivery Fraction (PDF), and routing overhead. The average ETED is the average time elapsed between when data packet was initially sent by the source node and when it was successfully received by the destination [4]. Routing overhead is the total number of RREQ packets generated and transmitted during the entire simulation period. For packet sent over multiple hops, each transmission over one hop is counted as one transmission [4]. PDF can be defined simply as the total number of data packets received by destinations as a fraction of the total number of data packets sent by sources [3]. We generated the proper scenarios and tracing files for required experiments. In the final stage we have made parsing for the generated tracing files using special parser in order to gain routing overhead, PDF, and average ETED values.

4.3.1. Average End-To-End Delay (ETED)

Figure 1 show the effect of different transmission range values on average ETED according to different probabilistic algorithms FP, 2P and 4P. From the

Figure 2: PDF vs. Transmission Range.

4.3.3. Routing Overhead

Figure 3 shows the effect of different transmission range values on routing overhead regarding to different probabilistic algorithms FP, 2P, and 4P. When the transmission range values and number of P increase the routing overhead values are decrease. From the figure, it is shown that there is a significant reduction in the number of RREQ that are sent when the transmission range is 300. This indicates that high tranmission range is perferable in such environments to reduce the RREQ ( route dicovery process overhead) packet overhead because when transmission range is small the number of hops needs to reaches a destination is more than when transmission range is large. Therefore, the number of RREQs that is used in route discovery process reduced.

Figure 3: Routing Overhead vs. Transmission Range.

5. Conclusion
Simple flooding is one of the mechanisms that are used to spread data over networks. This technique is suffering many challenges such as redundancy, contention and collision. Previous studies suggested solutions to this problem using probability flooding broadcast. Our work attend to show the behaviour of probability flooding approach FP , 2P and 4P with different transmission range values in terms of routing overhead, PDF, average end-end delay (ETED) using ns-2 simulator. The results show that when the transmission range values and number of P increases the performance of the multiple-Ps algorithms improved, where the protocol with higher P value (4P) achieve better results comparing to all other protocols in terms of PDF, ETED and the routing overhead.

References
[1] Bani Yassein Muneer, On the Effect of Mobility and Density on Probabilistic Flooding in MANETs , Proc. INT. Working Conference on Performance Modeling and Evaluation of Heterogeneous Networks

(HETNETs'04) , British Computer Society (BCS), IEEE, Ilkley, West Yorkshire, U.K, pp. 63/1- 63/9, 26th - 28th July, 2004. [2] Bani Yassein Muneer. A Performance Comparison of Smart Probabilistic Broadcasting of Ad hoc Distance Vector (AODV), 25th UK Performance Engineering Workshop Leeds,UK ,July, pp. 6-7, 2009 [3] Bani Yassein Muneer. Application of Probabilistic Flooding in MANETs. Ubiquitous Computing and Communication Jornal. February 2009. [4] Bani Yassein Muneer. On the Performance of Probabilistic Flooding in Mobile Ad Hoc Networks. Parallel and Distributed Systems, July 2005. [5] Bani Yassein Muneer. Smart Probabilistic Broadcasting in Mobile Ad Hoc Network. Mosharaka International Conference on Communication, Signal and Coding MIC-CSC 2008. [6] Bani-Yassein Muneer. Improving Route Discovery In On-Demand Routing Protocols Using Local Topology Information in MANETs, Proceedings of the Ninth ACM/IEEE International Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems (MSWIM 06), pp 95-99, October 2006. [7] Bani Yassein Muneer. Performance Evaluation of Adjusted Probabilistic Broadcasting in MANETs. Proceedings of the 2nd IEEE International Symposium on Dependable, Autonomic and Secure Computing (DASC'06), 2006. [8] Brad Williams, Comparison of broadcasting techniques for mobile ad hoc networks. Proceeding of ACM Symposium on Mobile Ad Hoc Networking & Computing (MOBIHOC 2002), pp. 194205, 2002 [9] C. Perkins. Ad hoc On-demand Distance Vector (AODV) routing. IETF Mobile Ad Hoc Networking Working Group, internet draft version 13, February 2003. [10] J. Cartigny. Border node transmission based probabilistic broadcast protocols in ad-hoc networks, Telecommunication Systems, vol. 22, no 1 4, pp. 189 204, 2003 [11] Julien C. Border Node Retransmission Based probabilistic Broadcast Protocol in Ad Hoc Networks. Proceedings of the 36th Annual Hawaii International Conference on System Sciences (HICSS03); 2003:293-303. [12] Nourddine Enneya. Enhancing AODV Performance based on Statistical Mobility Quantification. The IEEE International Conference on Information & Communication Technologies: from Theory to Applications - ICTTA06,pp. 2455 2460, April 2006. [13] Qi Z. Dynamic Probabilistic Broadcasting in MANETs. Journal of Parallel Distributed Computing. Elsevier; 2005; 65: 220-233. [14] S.-Y Ni. The broadcast storm problem in a mobile ad hoc network, ACM Wireless Networks, Vol. 8, No. 2, March 2002, pp. 153-167. (SCI, EI) [15] Sze-Yao N . The broadcast storm problem in a mobile ad hoc network. In the Proceedings of the 5th annual ACM/IEEE international conference on Mobile computing and networking (MobiCom '99). New York, NY, USA: ACM Press; 1999:151-162.

[16] Y. Sasson . Probabilistic broadcast for flooding in wireless mobile ad hoc networks, Proc. IEEE Wireless Communications & Networking Conference (WCNC 2003), pp. 1124-1130, March 2003. [17] Y. Zhang, Intrusion Detection in Wireless Ad-Hoc Networks, ACM MOBICOMM, pp. 275-283, 2000.