Applying trust enhancements to reactive routing protocols in mobile ad hoc networks

Abstract

Due to the characteristics of mobile ad hoc networks, such networks are more susceptible to the destruction of malicious attacks or denial of cooperation. It would be easy for an adversary or a malicious node to launch attacks on routing function, especially attacks on packet routing. In order to mitigate these hazards, we incorporate the concept of `trust' into MANETs, and abstract a decentralized trust inference model. The core of this model is trust computation, which is divided into two parts: historical trust assessment and trust prediction. We can quantify a node's historical trust based on its historical behaviors via introducing multiple trust attributes. The fuzzy AHP method based on entropy weights is used to calculate the weight of trust attributes. By making use of the obtained historical trust data sequence, we propose an improved dynamic grey-Markov chain prediction measure to effectively estimate node's trust prediction. In order to verify the validity of our trust model, we propose a trust-enhanced unicast routing protocol and a trust-enhanced multicast routing protocol, respectively. Both of the two new protocols can provide a feasible approach to kick out the untrustworthy nodes and choose the optimal trusted routing path. Moreover, the new proposed data-driven route maintenance mechanisms can reduce the routing overhead. The persuasive experiments have been conducted to evaluate the effectiveness of the new proposed trust-enhanced routing protocols in the aspects of packets delivery ratio, end-to-end latency, malicious node detection and attack resistance.