Probabilistic Mechanism to Avoid Broadcast Storm Problem in MANETS

, pp.479-486 http://dx.doi.org/1.14257/astl.217.147.67 Probabilistic Mechanism to Avoid Broadcast Storm Problem in MANETS G Parimala 1, B Suvarna 2, N Rajeswari 3 and Venkatesulu Dondeti 4 VFSTR University, Guntur, AP, India 1 garnepudi.parimala@gmail.com, 2 sv172@gmail.com, 3 rajeswari.nallamekala9@gmail.com, 4 drvenkatesulud43@gmail.com Abstract. The overhead of the route discovery can be increased due to frequent link breakages in the MANTES. This is caused by the high mobility of nodes in network. Broadcasting is the optimal mechanism for the route discovery in the MANETS. In this mobile node rebroadcasts the RREQ packet to all the neighbors blindly. By this the duplicated packets enormously arises in the network. This causes the broadcast storm problem. Here in our paper by combining the neighbor coverage based methods with the probabilistic methods to reduce the routing overhead in the network. By maintain some factors like rebroadcast delay, additional coverage ratio, and rebroadcast probability the duplicated packets in the network can be reduced. By combining the area based methods and the probability based methods the retransmissions can be reduced. By executing the NCPR routing performance can be increased and reduces the redundant rebroadcast compared to the AODV and DPR. Keywords: Broadcast storm problem, Rebroadcast, Neighbor-coverage methods, Routing overhead. 1 Introduction In wireless communication systems, maintaining connection between devices is not necessary because all the objects move freely throughout the network, This advantage creates some problems in the wireless communication systems. The most attractive characteristics of the MANETS are: Mobility: A wireless communication allows users to use the network resources from anywhere. The devices in the network can move freely from anywhere. Reach ability: In the mobile communication, the mobile nodes better connected to people which leads high reach ability. Simplicity: Wireless communication is much simpler than wired network. Initial setup of network is very easy compared to wired network. Maintainability: The maintainability of the mobile communication is bit difficult compared to the wired communication. The maintenance of the devices in the network should be under the network. Roaming services: As it is mobile communication, the services can be provided from anywhere including trains, busses etc. ISSN: 2287-1233 ASTL Copyright 217 SERSC

Wireless network has two types of communicating environments they are 1. Infrastructure based Ad-Hoc network. 2. Infrastructure-less Ad-Hoc networks. In Infrastructure based Ad-Hoc network, all the Aps (access points) in the network are connected to the fixed backbone and all the mobile nodes are free to move. Whereas in the infrastructure less network, there is no fixed backbone structure, no AP, only the mobile nodes are present in the network and move freely in the network. A Mobile nodes in MANET act as both end point and as a routers, as an end point it receives packet, as a router it forward the packets further. The mobile nodes are taking care of communication in the network. The communication among nodes is done by following 3 ways. First: route discovery from source to destination, second: establishing the optimal route third: forwarding the data packets from source to destination. Finding the optimal route between source and destination becomes major challenge in the MANETs. Various routing protocols are designed for finding the route from source and destination. Initially the routing protocols are of two types which are proactive and reactive routing protocols. In proactive routing the routing information is maintained at all the nodes as a table called as routing table from source to destination in the network. Whereas in reactive routing the routes in the network is found on-demand when it is necessary. In reactive routing type of routing mechanism routes are identified periodically and maintain the routes to every possible destination in the network. The main challenging issue in the MANET is found to be the route discovery. The route discovers phase is to find out all the routes from source to destination. In conventional on-demand routing protocol, for finding the route, the node should broadcast the Route Request packet (RREQ) packet to all the neighbors which are in that node s transmission range in the network. The received nodes simply (blindly) rebroadcast that RREQ packet until it discovers the route to the destination. This broadcasting is known as blind flooding. This process continues until destination is reached. At the end the total number of rebroadcast is N-2. Where N is the total number of nodes in the network. By this flooding mechanism most of the packets are going waste and bandwidth of the network is wasted by this forwarding of unnecessary packets. This can leads to excessive redundant retransmissions in the network. The collision of packets in the network is high and contention of data occurs. This serious contention, collision and redundant retransmissions lead to Broadcast storm problem. Many solutions are proposed to overcome this broadcast storm problem. 1.1 Broadcast Schemes Broadcasting is the most powerful mechanism that can be used to forward the packets in any mobile communication. As the mobility is very high in the MANETS, mobile nodes can move from one place to another in the network with respect to time. In this case broadcasting mechanism is used to forward packets. In this method node broadcasts the packets to all the nodes which are under the radio range of its node. That received nodes further rebroadcasts the packets to other neighbors. This process continues until the destination is reached. The uncovered neighbors are covered when 48 Copyright 217 SERSC

the rebroadcasting is takes place. For this broadcasting different techniques are used. Different broadcasting schemes are listed below: a. Unicasting: Forwarding message from one device to other device. b. Broadcasting: Forwarding message from one device to all the devices in the network. c. Multicasting: forwarding message from one device to specified devices. Basically the broadcasting schemes are classified into five types. They are: 1. Probabilistic based schemes: In this type of schemes, the RREQ packet is forwarded with some probability. Based on the probability, the transferring of packets are done. 2. Counter based schemes: In this counter based schemes, the RREQ packet is transferred based on time. That means, the sender rebroadcasts the packet when the time expires. In this type of counter based schemes the time factor is used. 3. Distance based schemes: In this distance based schemes, the distance between sender and receiver can be measured. Based on the distance the packet forwarding is takes place. If the distance between sender and receiver. When the distance between sender and the receiver is high then the rebroadcasting probability is high. 4. Location based schemes: In this type of location based schemes, the location of the node can be calculated by Global positioning system (GPS). Based on that information, the forwarding of the RREQ packets is done. 5. Cluster based schemes: In this scheme, some of mobile nodes form a cluster and communicate with the other clusters in the network. By introducing this type of methods bandwidth consumption is less. And the active members in cluster based schemes are the cluster heads and gateways. The broadcasting protocols used traditionally are: 1. Flooding Different approaches are proposed for the broadcasting in the mobile communications. But there are some problems in those methods. Flooding is the optimal mechanism for the broadcasting. In this flooding, one node floods the RREQ packets to all the nodes in the network. That means the main aim of this flooding mechanism is to cover all the nodes. The process in flooding is one node sends the RREQ packet to the nodes which are in the radio range. Then it is the responsibility of the receiving node to forward the RREQ packets further in the network. By this packet collisions are more. And the bandwidth of the network is wasted. By this flooding mechanism is not optimal one for the broadcasting in the MANETS. Some improvements are made for this flooding mechanism. 2. Probabilistic broadcasting scheme based on coverage area & neighbor confirmation: This broadcasting is developed by using some other techniques; by appending probability to the broadcasting some of the problems with the broadcasting can be reduced. In this approach we use a dynamic probabilistic approach based on the coverage area and neighbor confirmation. In this mechanism, coverage area concept is used. If the mobile node is placed near to the sender, then the rebroadcast probability is low. Whereas the mobile node is placed far from sender then the rebroadcast Copyright 217 SERSC 481

probability is high. In this case the additional coverage area is high. The coverage area of one broadcast is calculated by measuring the distance between the sender and receiver. The node position can be measured by global position system (GPS). Broadcasting is done based on some probability of the RREQ packet. The main aim of this scheme is to reduce the number of rebroadcasts; by this the bandwidth of the network is not wasted. By this dynamic based probabilistic broadcasting scheme the rebroadcast probability is less and the network resources are not wasted compared to the previous method. 3. Scalable broadcast algorithm: In this algorithm the main aim is to reduce the cost of the rebroadcast. Mobile node need not to rebroadcast when all the neighbors is covered by this broadcast. In this algorithm is divided in two phases. First it needs to find out the neighbors list of the node. By periodically sending hello packets, each and every node maintains the neighbor list. Each and every node maintains the topology information of 2 hop neighbors. When broadcasting is takes place, then the receiver checks which neighbors are covered by this broadcast by checking into the neighbor information. Then those covered nodes are added into the broadcast coverage nodes list. Then any rebroadcast decision is made, it checks into that list. If all its neighbors are covered, the rebroadcast id canceled. By this Scalable broadcast algorithm, the redundancy of rebroadcast is decreases. It saves the duplicate packets by compared to the flooding. 4. Probabilistic schemes by calculating delay This NCPR is the protocol for the routing in the network which has minimum routing overhead. This NCPR requires one-hop neighbors of the node. Some parameters used in this NCPR, which are rebroadcast delay, uncovered neighbor sets, rebroadcast probability. The above methods have less overhead compared to the flooding. But in the case of number of nodes are high then the overhead is also high. When the routing is started, first the source node sends the RREQ packet to the nodes which are in the transmission range of that source node. Then the nodes which received that RREQ packets need to resend the RREQ packets further to the nodes until it reaches the destination node. In this process the packet overhead is increased. 1. NCPR In this NCPR first the mobile need to identify the 1-Hop neighbors. There after it calculates the uncover neighbors list. a. Uncovered neighbor list: First the mobile node sends the RREQ packet to the all 1-Hop neighbors. After that it calculates the uncovered neighbors list of that broadcast. UC(a) = N(a) [N(a) N(b)] {b} Where UC(a) is the Uncovered neighbors of the node a. N(a) is the neighbors set of the node a. N(b) is the neighbor set of the node b. here node b is the source node and the node a is the neighbor of the node b. b. Rebroadcast delay: By calculating this rebroadcast delay, the routing overhead can be reduced. If the node a has more common neighbors uncovered by the RREQ packet from the node b. then the delay of that node is low. If the packet is rebroadcasted through the lower delayed node, then the more common neighbors can know the RREQ packet in the network. The rebroadcast delay of the node can be defined as follows: 482 Copyright 217 SERSC

N(b) N(a) D r = 1 N(b) D = Max. delay D r Where D r is the delay ratio of the neighbor node. N(b) is the neighbor list of the source node. N(a) is the neighbor list of node a. and D is the rebroadcast delay of the node neighbor node. Max.Delay is the smallest constant delay. c. Additional coverage ratio: For the rebroadcast delay timer should be set. If the timer expires then the final uncovered neighbor set of the RREQ packet finalized by the node. After getting the final uncovered neighbors set, then the rebroadcast probability of the node should be calculated. For this the additional coverage ratio of the node should be defined. Additional coverage ratio is the ratio of the number of nodes that are additionally covered by this broadcast to the total number of nodes in the network. R(a) = UN(a) N(a) Where R(a) is the additional covered ratio. UN(a) is the uncovered neighbors of node a. N(a) is the neighbors set of node a. d. Connectivity factor The connectivity of the network should reach 1 at any time. To keep this the measure which is connectivity factor should be maintained. In the [16] reference the connectivity metric is derived as 5.1774 log n. Where n is the number of nodes in the network. cf(a) = N c N(a) Where cf(a) is the connectivity factor of the neighbor node. N c is the 5.1774 log n. here in this connectivity factor the denominator is greater than the numerator then the cf(a) is less than 1. That means node a is in the dense area of the network. The part of neighbors of node a sends the RREQ packets further. Then only the network connectivity keeps constant. Otherwise cf(a is ) greater than 1. That means the node is the sparse are of the network. Then the node a should send the RREQ packets to keep the connectivity factor. e. Rebroadcast probability By combining the additional coverage ratio and connectivity factor the rebroadcast probability of node can be obtained. By this rebroadcast probability the decision is made whether the RREQ packet is further rebroadcast or not. RBP = R(a) cf(a) 2 Implementation of NCPR To implement the proposed protocol the AODV code is modified to reduce the routing overhead in the network. To evaluate the performance of the proposed protocol, it should be compared with some other protocols in NS2. DPR is the optimal mechanism for the RREQ packet broadcasting in MANETS. This can be compared with the NCPR. The simulation is started in the 15 15 area. The number of Copyright 217 SERSC 483

Advanced Science and Technology Letters nodes in the network can be any number. The Max. Delay in the above formula should be set to.1 s. The performance of the routing protocol can be measured based on following metrics. Normalized Routing overhead MAC collision rate: The number of packets dropped at MAC layer. Routing overhead: The routing overhead when the packet is transferring form source to destination. Packet delivery ratio: The ratio of number of data packets successfully received by the destination to the number of packets generated by the source. End-to-End delay: Average delay of successfully delivered packets from source to the destination. 2.5 2 1.5 1.5 NCPR DPR 5 1 15 2 25 3 35 Number of nodes Fig. 1. Routing overhead with varied number of nodes Packet delivery ratio% The NCPR protocol reduces the routing overhead in case of large number of nodes during route discovery phase. 1 95 9 85 8 75 7 65 NCPR DPR AODV 1 2 3 4 Number of nodes Fig. 2. Packet delivery ratio with varied number of nodes NCPR protocol can increase the packet delivery ratio with varied number of nodes because it reduces the collisions in the network. 484 Copyright 217 SERSC

Advanced Science and Technology Letters Average end to end delay.8 NCPR DPR AODV.6.4.2 5 1 15 2 25 3 35 Number of nodes Fig. 3. End-to-End delay with varied number of nodes The NCPR protocol decreases the end-to-end delay with varied number of node compared to the AODV and DPR. Normalized routing over head 1.5 NCPR 1 DPR.5 1 12 14 16 18 2 Number of CBR connections Fig. 4. Routing overhead with varied number of CBR connections NCPR reduces the routing overhead with number of CBR connections increased compared to the AODV and DPR. 4 Conclusion In this paper the proposed NCPR protocol works optimal when compared to the AODV and other broadcasting schemes. NCPR reduces the routing overhead with varied number of nodes. The broadcast storm problem can be reduced by calculating rebroadcast delay. By that the broadcasting order can be calculated. By combining additional coverage ratio and connectivity factor the rebroadcast probability cane be calculated. According to that rebroadcast probability, sending of the packet further can be decided. The NCPR protocol reduces the retransmissions compared to the AODV. The simulation results show that the proposed protocol reduces the routing overhead when the network density is high. Copyright 217 SERSC 485

4 Future Enhancement NCPR reduces the routing overhead in the network. And the broadcast storm problem can be resolved by this NCPR. In future this NCPR is designed to reduce the energy consumption. References 1. S.Y. Ni, Y.C. Tseng, Y.S. Chen, and J.P. Sheu, The Broadcast Storm Problem in a Mobile Ad Hoc Network, Proc. ACM/IEEE MobiCom, pp. 151-162, 1999. 2. B. Williams and T. Camp, Comparison of Broadcasting Techniques for Mobile Ad Hoc Networks, Proc. ACM MobiHoc, pp. 194-25, 22. 3. C. Perkins, E. Belding-Royer, and S. Das, Ad Hoc On-DemandDistance Vector (AODV) Routing, IETF RFC 3561, 23. 4. D. Johnson, Y. Hu, and D. Maltz, The Dynamic Source Routing Protocol for Mobile Ad Hoc Networks (DSR) for IPv4, IETF RFC 4728, vol. 15, pp. 153-181, 27. 5. J.D. Abdulai, M. Ould-Khaoua, L.M. Mackenzie, and A. Mohammed, Neighbour Coverage: A Dynamic Probabilistic Route Discovery for Mobile Ad Hoc Networks, Proc. Int l Symp. Performance Evaluation of Computer and Telecomm. Systems (SPECTS8), pp. 165-172, 28. 6. R. Bar-Yehuda,. Goldreich, and A. Itai. Efficient emulation of single-hop radio network with collision detection on multi-hop radio network with no collision detection. Distributed Computing, 5(2):67-72, 1991. 7. J. Broth, D. A. Maltz, D. B. Johnson, Y.-C. Hu, and J. Jetcheva. A performance comparison of multi-hop wireless ad hoc network routing protocols. In Proc. IEEE/ACM Intl. Co@ on Mobile Computing and Networking MOBICOM, pages 85-97, 1998. 8. J. Broth, D. A. Maltz, D. B. Johnson, Y.-C. Hu, and J. Jetcheva. A performance comparison of multi-hop wireless ad hoc network routing protocols. In Proc. IEEE/ACM Intl. Co@ on Mobile Computing and Networking MOBICOM, pages 85-97, 1998. 9. J.D. Abdulai, M. Ould-Khaoua, and L.M. Mackenzie, Improving Probabilistic Route Discovery in Mobile Ad Hoc Networks, Proc. IEEE Conf. Local Computer Networks, pp. 739-746, 27. 486 Copyright 217 SERSC