JOURNAL OF COMPUTING, VOLUME 3, ISSUE 7, JULY 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.

ORG

102

A Multi-path Misbehaviour Aware Routing Scheme for Mobile Ad Hoc Networks

Mr. Bhushan M. Manjre, Mrs.Veena A. Gulhane
AbstractIn a M ANET, Node misbehavior is any such behavior that proves harmful to co-operative environment of MANET. Many schemes have been recently proposed for the detection and avoidance of misbehaving nodes, but still there are many issues like false detection due t o network layer factors, packet dropping, and packet delaying which are yet to be addressed completely. To mitigate these problems, a novel multi-path routing protocol is proposed herewith that aims at finding reliable and secure routes for data communication before transmitting data packets over the same. Since routes are verified for security and reliability at the beginning of data transmission, hence probability of packet loss/delay misbehavior is minimized.It also implements behavior check mechanism over failure routes to point out the culprit node, checks whether it is misbehaving due to some network layer factors or it is intentionally programmed to do so. If it is intentionally misbehaving, then only it is avoided for future routes, otherwise not. Hence, false detection due to low energy and buffer overflow is avoided and the loyal node is given chance of reintroduction into the network again. Due to this, the network performance is optimized in terms of security and reliability. The proposed protocol, when compared with AOMDV protocol, shows improvement in terms of network throughput and reduction in end to end delay.

Index TermsCo-operative multi hop forwarding capability, receiver collision, path efficiency, misbehavior etc.

1 INTRODUCTION
Mobile Ad Hoc Network is a set of mobile nodes that form wireless network without any fixed infrastructure. Each node pays role of both, a host, i.e an end system that executes applications and acts as source or destination and secondly router that relays data traffic for randomly, hence network topology alters frequently. The network topology depends upon the current location and transmitting power of nodes. When node acts as router, its main task is to forward data packets for other nodes plus discovery and maintenance of routes to the destination. There are two types of routing protocols: Proactive routing protocol and Reactive routing protocol. Proactive routing protocol: It stores and maintains route to destination by periodically updating with the help of control overhead. Reactive routing protocol: In this type, route computation is made only when it is needed. One of the biggest challenges in ad-hoc networking is the efficient delivery of data packets from source to destination especially in the situation where each device in a MANET is free to move independently in any direction with any speed, and will therefore frequently changing its links to other devices and breaking current paths. Hence, due to the wireless and distributed nature of MANETS, routing in ad-hoc networks can be viewed as a challenge. In such dynamic network, it is important to get route in time, perform the routing that gives maximal throughput with minimal control overhead since control messages consume both channel bandwidth as well as the battery power of nodes for Communication/processing. Mobile ad hoc networks are a wireless network in which paths between sources to destination are formed on ad hoc basis and each intermediate node in the path has to act as router and forward packets between source and destination. In such self-organized networks, each node has to forward data traffic unrelated to its own use. But being a router for other nodes leads to consumption of battery, processing and bandwidth resources of the router node. So, in order to achieve maximum throughput with the available resources, a node may not be willing to contribute their resources to maintain network connectivity to save its resources. Such selfish behavior may result into damages like denial of service which in turns degrades the performance of the network in terms of network throughput and packet delivery ratio because most existing routing protocols in MANET are aiming at finding most efficient path. There are two main security issues with ad hoc networks [17]. The first is the need for privacy of communication in network where communication transmission is being performed by nodes owned by many different people/organizations. The other is that the network is vulnerable to a number of attacks (not necessarily deliberately) that can degrade the performance of the network or give unfair advantage to some of the participants. Due to lack of preexisting infrastructure, it is easier for attackers to enter or leave network, perform eavesdropping and gain access to confiden Mr. Bhushan M. Manjre is student of G.H. Raisoni College of Engineertial information since no physical connection is needed for it. ing, Nagpur, India. As there is no online certification authority or trusted third Mrs. Veena A. Gulhane is Assistant Proffesor in Department of Computer Science and Engineering, G.H.Raisoni College of Engineering, Nagpur, party and hence security concerns are more complicated in
India.

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 7, JULY 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

103

MANETS as compare to conventional networks. The other main issue is that the functions of the network are provided by the user devices themselves rather than an independent network operator [17]. It means there is no governing body that can compel nodes to co-operate each other. Also the user devices can be programmed like mobile phones, hence they can be easily made to behave in any desirable way. In future, ill minded users may program their own devices to achieve desirable ill effects.

1.1 Misbehaviour of Nodes Misbehavior of Node is any such behavior that goes in total conflict of cooperative working environment of an ad hoc network. A misbehavior threat can be defined as an unauthorized behavior of an internal node that can result in unintentional damage to other nodes, i.e., the aim of the node may not to launch an attack, but it may have other aims such as obtaining an unfair advantage compared with the other nodes [17]. Nodes will misbehave if controlled or programmed to do so by their owners or users with distinct dimensions of misbehavior as follows [15]: a) Accidental or deliberate.
b) Selfish or malicious. c) Individual or collusion. Hence if we can identify and avoid misbehaving nodes during communication session, we can prevent the overall operation of ad hoc networks from getting hampered from various perspectives. In this paper, a novel multi path routing scheme is proposed to address the above mentioned routing layer misbehaviour.

Enhanced Ad-Hoc on Demand Multi path Distance Vector Routing Protocol (EAOMDV) [4] was proposed to address route failure problem in AOMDV by pre-emptively predicting the link failure by the signal power received by the receiver. In the Detection of Colluding Misbehaving Nodes in Mobile Ad hoc and Wireless Mesh Networks approach[18], authors propose Leak Detection Mechanism to address colluding node misbehaviour. The main idea of Leak Detector is that the destination node of a route builds up a virtual graph, which models the multi path from the source node to the destination node. Periodic traffic information (which can be piggybacked on the proactive routing messages) enables the destination node to calculate the ratio of incoming and outgoing traffic corresponding to the multi path routing information for each participating node. Using graph theory, traffic leaks are identified. In the Improved AOMDV with Precaution Algorithm in Wireless Ad Hoc Networking approach[12], in order to improve the performance of AOMDV, a precaution mechanism is introduced by authors to ensure the validity of routing information stored in each node and reduce the unnecessary control overhead in the route finding procedure. The precaution mechanism forecasts the possible failures of the links by evaluating two characteristics: the density of nodes and the overlapping of routes. In the paper titled Performance Evaluation and Comparison of AODV and AOMDV[3], authors have compared and evaluate the performance of two types of on demand routing protocols- Ad-hoc On-demand Distance Vector (AODV) routing protocol, which is uni path and Ad hoc On-demand Multi path Distance Vector (AOMDV) routing protocol. In this paper they note that on comparing the performance of AODV and AOMDV, AOMDV incurs more routing overhead and packet delay than AODV but it had a better efficiency when it comes to number of packets dropped and packet delivery.

1.2 Organization The rest of the paper is organized as follows: Section 2 describes the related work. Section 3 describes proposed work. Section 4 shows simulation results. Section 5 concludes this paper and outlines the future work. Section 6 points out references.

3. PROPOSED WORK 2 RELATED WORK

Security in MANET has been an active research area. Many solutions for detection and avoidance of misbehaving nodes have been suggested in the literature. A Distributed Cooperative Approach [15] is proposed to improve detection and removal of misbehaving MANET Nodes in which issue of misbehaving nodes is addressed by providing a distributed cooperative system, in which every node participates in identifying the misbehaving node. Every node exchanges its monitored information both cooperative as well as non-cooperative and giving chance of nodes reintroduction into the network in case of false detection. Multi path Reliable Routing (MRR) algorithm [14] determines reliable paths based on node reliability index parameter and adds redundancy in transmitted data to reduce data loss. The proposed mechanism is a model of secure and reliable multi-path reactive routing protocol for mobile ad hoc networks. It is divided into three modules in order to facilitate its analysis. Module 1 comprises of identification of misbehaving nodes. Module 2 will remove the threats imposed by misbehaving nodes. Module 3 explains optimization of network performance. The modus operandi of these modules is explained in detail as follows:

3.1 Detection of Misbehaving Nodes

Proposed routing protocol uses an Multiple Routes Set (MRS) comprising node-disjoint paths, determined using the AOMDV protocol .Each loyal node where proposed protocol is installed has been given a matrix of dimension 55. During the route discovery, each node asks a unique question to its neighbour node. If the neighbour node answers it

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 7, JULY 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

104

correctly then only that node is entertained in route discovery otherwise not. Every node has same matrix of dimension 55. The question is formed by a random row and random column number of matrix and asked to neighbour node. Suppose the question is 21, it means 2nd row and 1st column. Neighbour node retrieves the value at 2nd row and 1st column in its matrix and gives that value back to the node who asked the question. The questioning node retrieves the value at 2nd row and 1st column in its matrix and if it is matched with value given by answering node, it means that answering node is not any external attacker and can be trusted for route discovery phase. Multiple routes between two nodes can be either linkdisjoint or node disjoint. In the node disjoint method, nodes on the routes should not be common, whereas, in the link disjoint method, links on the routes should not be common [19]. Thus, traffic load on the shared node in link disjoint route will be much higher than the other nodes of the routes. As a result, this node tends to die much earlier than the other nodes, leading to the routes to break down much earlier. Thus, the presence of node disjoint routes prolongs the network lifetime by reducing the energy depletion rate of a specific node [3]. In the proposed protocol, multiple routes are used for sending data from source to destination and thus making data transfer faster and achieving load balancing in situations where the data traffic is heavy. Now, once the MRS is filled with node disjoint routes, a unique id is assigned to each node disjoint route from 0 to n. Here two types of control packets viz. TPI (Total path information) Packet and PFI (Path failure information) packet are used. TPI packet is consisting of five fields as follows: 1) Total number of routes obtained in route discovery i.e. n. 2) The route id of the route over which that particular TPI packet is sent. 3) Source node Id 4) Destination node Id 5) Timeout value which is the time threshold for which the destination will wait for TPI packets to arrive at it. PFI Packet contains following fields: 1) Ids of failure routes 2) Alert identifiers for failure routes. This alert identifier indicates type of misbehaviour i.e. its value is 0 if node misbehaviour is packet delaying, and if it is packet dropping then its value is 1. The overall mechanism to identify misbehaving nodes works as follows: Initially the TPI packets are sent over all routes in MRS, one TPI packet per route. Each TPI packet contains total count of the routes obtained in route discovery. Whenever first TPI packet reaches to destination, the destination node extracts it to obtain two important values i.e. total number of routes and timeout value. The value total number of routes gives destination, the total count of all node disjoint routes obtained during route discovery, between source node and destination node, over which it is supposed to get TPI packets. The second value i.e. timeout value gives destination, the time threshold for which the destination node should wait for TPI packets to arrive at it. If any TPI packet reaches to destination after this timeout value, then it is concluded that

the route over which that TPI packet arrived, contains misbehaving node who is delaying the packets. Now when destination obtains any TPI packet, it extracts that packet to get the route id over which it has arrived and records that route id. If any TPI control packet is lost or delayed in middle of its route, then corresponding route id will not be reported to the destination within particular time threshold, which means that the route over which that TPI packet is suppose to arrive, has dropped or delayed it. After waiting for particular time threshold i.e. timeout value, destination sends PFI packet to source containing the ids of failure routes and the alert identifiers, over the shortest route chosen among the routes in MRS. As soon the source obtains PFI packet, it starts data transmission over all routes except the failure routes reported by PFI packet. At the same time, it triggers the behavior check mechanism over failure routes, one by one. Stepwise summary of the above mechanism is as follows: Step 1) Route Discovery for Node-Disjoint Multiple Routes Set (MRS) using AOMDV protocol, inclusive of questionanswering session. Step 2) Forwarding TPI packets over all routes in MRS. Step 3) Destination checks for missing TPI control packets and records the ids of failure routes. Step 4) Sends PFI packet back to source containing the ids of failure routes. Step 5) Source will avoid failure routes reported by PFI packets.

3.2 Removal of Misbehaving Nodes

As soon the source node receives PFI packet then in addition to commencement of data transmission, it also trigger behaviour check mechanism that will check failure node in the failure route for two conditions as follows: Whether its energy is depleted. Whether its buffer has overflowed. If its energy is depleted or its buffer is overflowed, it means that it is loyal node but because of the above reasons, it is not able to forward the data packets. Such nodes are avoided temporarily but not blacklisted. But if neither the energy of node is depleted, nor its buffer is overflowed, then the node is declared as misbehaving node, which intentionally programmed to misbehave and thus it is blacklisted. As the behaviour check mechanism points out packet dropper/delaying node it informs about misbehaving nodes to source. Source will put misbehaving node in the blacklist maintained at source. Those nodes which are blacklisted are avoided in next route discovery. By doing this, their presence in future routes is eliminated. Now the MRS has reliable routes. But it is also possible that any node in the reliable route may starts misbehaving in the middle of communication. In such cases, source wont be getting acknowledgement

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 7, JULY 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

105

for dropped packet within retransmission time out (RTO). Here source will point out failure route from routing table, stops further data transmission over the same. Also, it redirects traffic of failure route over next available route and triggers behaviour check mechanism over failure route so as to check this route for misbehaving reasons and to blacklist the misbehaving nodes if any. The packet that was dropped over this failure route is retransmitted over new route which was recently selected for data transmission so as to avoid packet loss. The mechanism for eliminating misbehaving nodes is summarized in steps as follows: Step 1) Source triggers behaviour check mechanism over failure route, to check the node due to which the route is failed. Step 2) The node is investigated for the reasons behind misbehaviour. If the node is intentionally misbehaving, then only it is blacklisted. Step 3) Source maintains black list of misbehaving nodes and will exclude those in next route discovery. Step 4) If node in reliable route starts misbehaving in the middle, then source wont be getting its acknowledgment within retransmission time out (RTO). Step 5) Source will check its routing table, point outs missing packet route and triggers behaviour checking mechanism over it and once the misbehaving node is detected, it is added in blacklist and will be avoided during next route discovery. Step 7) The missing packet is sent over another reliable shortest route chosen for further data transmission, from MRS so as to minimize packet loss and further data transmission continues.

mized. This is summarized in steps as follows: Step 1) Avoidance of False Detection: False detection due to low energy and buffer overflow is avoided. Hence it gives chance of reintroduction into the network to those loyal nodes which are not able to forward the packets temporarily due to network layer factors. Step 2) Avoidance of Reputation Based System: Since Reputation Based System is based on promiscuous overhearing which induces computational complexity, control overhead and takes lots of processing power and induces delay; hence by avoiding this, network performance is accelerated.

4. SIMULATION RESULTS

3.3 Network Optimization

There may be packet dropping because of several reasons like low energy and buffer overflow. Hence even though the node is not misbehaving intentionally still it is declared as packet dropping/delaying node and this leads to false detection. Due to false detection, reliable nodes are ignored which minimizes the total number of nodes taking part in data communication [21]. This result into less number of routes or long routes obtained in route discovery and thus may degrade the overall performance of the Network. The proposed protocol avoids this degradation due to false detection with the help of behaviour check mechanism. Secondly the routes are checked at the beginning of the data transmission by dispersing TPI packets and then sending data packets over it. In the first round itself, reputation of the node is identified. Thus, there is no need to employ reputation base system where the reputation index of the node is calculated throughout the promiscuous overhearing [20]. Hence computational complexity, control overhead, consumption of processing power, and excessive delay is mini-

fig.1: Delay Graph The fig.1 depicts the comparison of standard AOMDV protocol with the proposed multi-path routing protocol in terms of end to end delay. On X axis, number of packets are shown and Y axis shows delay in seconds. The red line indicates delay in proposed protocol and green line indicates delay in AOMDV protocol. Fig.2 shows the comparison of standard AOMDV protocol with the proposed multi-path routing protocol in terms of network throughput. On X axis, simulation time is shown and Y axis shows network throughput. The red line indicates network throughput obtained by proposed protocol and green line indicates network throughput obtained by AOMDV protocol.

JOURNAL OF COMPUTING, VOLUME 3, ISSUE 7, JULY 2011, ISSN 2151-9617 HTTPS://SITES.GOOGLE.COM/SITE/JOURNALOFCOMPUTING/ WWW.JOURNALOFCOMPUTING.ORG

106

fig.2: Throughput Graph

5. CONCLUSION
From the fig. 1 obtained for Delay, it can be observed that the end to end delay i.e. average time taken by packets to reach from source to destination is considerably reduced with proposed protocol, when compared with AOMDV. It is because, whenever packet fails to reach to destination, it is informed to source within a particular time threshold. This time threshold is greater in AOMDV. Hence, in AOMDV, the delay incurred is more. From the fig.2 obtained for Throughput, when the proposed approach is compared with AOMDV, it shows that the proposed multi-path routing protocol gives considerable throughput improvement. It is because, whenever there packets are dropped, the most of missing packets are retransmitted again over some another reliable route. Hence packet loss is kept minimum. The proposed approach avoids false detection and computational complexity and thus makes the overall network operation robust.

There is one more dimension of misbehavior, i.e. packet altering misbehavior where the packet contents are altered. This dimension can be addressed by implementing encryption and decryption or hashing technique and it is the future scope of the proposed protocol.

6. REFERENCES
[1] Sintayehu Dehnie and Stefano Tomasin Detection of Selsh Nodes in Networks Using CoopMAC Protocol with ARQ, IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL, . 9, NO. 7, JULY 2010 . Fahad T. Bin Muhaya , Fazl-e-Hadi, AtifNaseer Selfish Node Detection in Wireless Mesh Networks, International Conference on Networking and Information Technology 2010. S. R. Biradar, Koushik Majumder, Subir Kumar Sarkar, Performance Evaluation and Comparison of AODV and AOMDV, S.R.Biradar et al. / (IJCSE) International Journal on Computer Science and Engineering , Vol. 02, No. 02, 2010, 373-

[2]

[3]

377. Mrs.Sujata V.Mallapur, Prof. Sujata .Terdal Enhanced Ad-Hoc on Demand Multipath Distance Vector Routing Potocol (EAOMDV) ,(IJCSIS) International Journal of Computer Science and Information Security, Vol. 07 No. 03 March 2010. [5] Souene Djahel, Farid Nait-abdesselam, and Zonghua Zhang, Mitigating Packet Dropping Problem in Mobile AdHoc Networks: Proposals and Challenges, IEEE COMMUNICATIONS SURVEYS & TUTORIALS, ACCEPTED FOR PUBLICATION, IEEE 2010. [6] Mustapha Reda Senouci, Abdelouahid Derhab, Nadjib Badache , Efcient Monitoring Mechanisms for Cooperative Storage in Mobile Ad-Hoc Networks: Detection Time and Accuracy Tradeoffs, 2009 29th IEEE International Conference on Distributed Computing Systems Workshops. [7] Sintayehu Dehnie, Stefano Tomasin , Reza Ghanadan, Sequential Detection ofMisbehaving Nodes in Cooperative Networks with HARQ, 2009 IEEE. [8] Nastooh Taheri Javan, Reza Kiaeifar, Bahram Hakhamaneshi, Mehdi Dehghan, ZD-AOMDV: A New Routing Algorithm for Mobile Ad-Hoc Networks, 2009 Eigth IEEE/ACIS International Conference on Computer and Information Science. [9] Wenjia Li, Anupam Joshi, and Tim Finin, Policy-based Malicious Peer Detection in Ad Hoc Networks ,2009 International Conference on Computational Science and Engineering. [10] Wei Gong, Zhiyang You,, Danning Chen, Xibin Zhao, Ming Gu, Kwok-Yan Lam, Trust Based Malicious Nodes Detection in MANET, IEEE 2009. [11] K. Gopalakrishnan and V. Rhymend Uthariaraj, Scenario Based Evaluation of the Impact of Misbehaving Nodes in Mobile Ad hoc Networks, ICAC 2009. [12] Yuanqing Zheng, Zhiwen Zhao, Jiaojiao Zhu, Improved AOMDV with Precaution Algorithm in Wireless Ad Hoc Networking ,2009 IEEE International Conference on Control and Automation,Christchurch, New Zealand, December 9-11, 2009. [13] Dinesh Mishra, Yogendra Kumar Jain, Sudhir Agrawal, Behavior Analysis of Malicious Node in the Different Routing Algorithms in Mobile Ad Hoc Network (MANET)., 2009 International Conference on Advances in Computing, Control, and Telecommunication Technologies. [14] Dhanalakshmi Somasundaram and Dr. Rajaram Marimuthu, A Multipath Reliable Routing for Detection and Isolation of Malicious Nodes in MANET, 2008 International Conference on Computing, Communication and Networking (ICCCN 2008). [15] A. Dadhich, Dr. A. K. Sarje, Dr. (Mrs.) K. Garg , Distributed Cooperative Approach to improve detection and removal of misbehaving MANET Nodes,IEEE2008. [16] Jyotirmoy Karjee, Sudipta Banerjee, Tracing the Abnormal Behavior of Malicious Nodes in MANET,IEEE 2008. [17] Peter Smyth, Mobile and wireless communications: key technologies and future applications (London, UK: IEE, 2004). [18] Kalman Graf, Parag S. Mogre, Matthias Hollick, and Ralf Steinmetz, Detection of Colluding Misbehaving Nodes in Mobile Ad hoc and Wireless Mesh Networks, IEEE 2007. [19] Zhuo Lu, Cliff Wang and Wenye Wang, On the Impact of Backoff Misbehaving Nodes in IEEE 802.11 Networks, IEEE ICC 2010. [20] Wenjia Li, Anupam Joshi (IEEE Senior Member), and Tim Finin, Coping with Node Misbehaviors in Ad Hoc Networks: A Multi-Dimensional Trust Management Approach, Eleventh International Conference on Mobile Data Management, 2010 IEEE. [21] Hongxun Liu, Jose G. Delgado-Frias, and Sirisha Medidi, Using a Cache Scheme to Detect Misbehaving Nodes in Mobile Ad-Hoc Networks, ICON 2007 IEEE. [4]