Reputation-based cooperative detection model of selfish nodes in cluster-based QoS-OLSR protocol. (c2014)

Abstract

The QOLSR is a multimedia protocol that was designed on top of the optimized link state routing protocol for mobile ad hoc network. It considers the quality of service (QoS) of nodes in the network when selecting the multi-point relay (MPRs) nodes. This proto-col suffer major drawbacks regarding network lifetime, where nodes with high bandwidth but limited energy can be selected to serve as MPRs. This would drain the nodes resid-ual energy and shorten the network lifetime, and increase selfish nodes that degrade the network lifetime. The limited energy and resources, and the absence of any moti-vation mechanism cause mobile nodes to act selfishly when selecting the MPRs. In this thesis, we consider the tradeoff between prolonging the ad hoc network lifetime and QoS assurance based on QOLSR routing protocol. This can be attained by (1) decreasing the Multi-Point Relay (MPR) nodes without sacrificing the QoS and (2) taking into consideration the residual energy level, connectivity index, and bandwidth of these relay nodes. The mentioned goals can be attained by implementing the clustering model to QOLSR. Therefore, we suggest a new clustering algorithm and a MPR node selection based on different combinations of metrics, such as connectivity, residual energy, and bandwidth. Moreover, we consider the selfishness during the election and selection process by proposing the use of reputation system that will motivate nodes to participate during the selection of MPRs, where the reputation is calculated based on VCG mechanism design. After solving the selfishness during network formation, we have discovered that nodes can misbehave after being selected/elected. Such a passive malicious behavior could lead to a denial of service attack due to the drop of packets. As a solution, we propose a hierarchal cooperative watchdog detection model for the cluster-based QOLSR, where nodes cooperate in a hierarchical manner to detect selfish n odes. Furthermore, to motivate watchdogs to monitor and cooperate with each other, incentives are given and calculated using cooperative game theory, where Shapley value is used to compute the contribution of each watchdog on the final decision. Simulation results show that the novel cluster-based QoS-OLSR model can well extend the network lifetime, ensure QoS and decrease delay. Adding reputation as one of the QoS metrics, motivates nodes to act normally without sacrificing the quality of service of the network. In addition, the hierarchical cooperative detection model shows a more reliable and efficient detection of selfish nodes.