A Trust Based Dual Reinforcement Q-Routing protocol to detect misbehaving nodes in WMNs

Abstract

Nowadays, Wireless Mesh Networks (WMN) owing to its self adaptive nature are popular and used for diverse applications like community networks, broadband wireless access, E-healthcare etc.,. Secured routing in WMNs using cryptography, authentication is extensively addressed by researchers and the effects of misbehaving nodes that leads to the performance degradation of the network are explored mostly. The trustworthiness of every individual participating node while routing is essential to enhance the security of WMN. In this paper, a trust based dual reinforcement Q-Routing algorithm is proposed to detect the nodes performing packet dropping attacks in WMNs using cross layer approach. The Q-Routing algorithm uses reinforcement learning technique to detect the best path to reach the destination with minimal delivery time. But it is vulnerable to packet dropping attacks accomplished by misbehaving nodes in the network. The proposed model uses two trust components namely, behavioral trust and implied trust to compute the trust of the neighbor nodes by monitoring the behavior of nodes and links while forwarding the packets. The trust value of a link between the nodes computed at the data link layer and the trust value of the intermediate node computed at the network layer are the two cross layer metrics used by the proposed algorithm. These trust values are used to compute Q-Value of the nodes in addition to the minimal delivery time. Packet drops due to buffer overflow and energy depletion is also detected by this method which reduces false positive rate. The effect of packet dropping attacks on Q-Routing and the proposed Trust Based Dual Reinforcement Q-Routing (TBDQR) are analyzed using simulation. The experimental results confirm that TBDQR detects the packet dropping nodes and results in an increased packet delivery ratio compared with traditional Q-Routing.