A cooperative recommendation trust model for ad hoc networks

Abstract

Since traditional security methods do not work very well in ad hoc networks, trust frameworks are introduced as a supplement to security mechanism. In current trust evaluation systems, direct trust information and recommendation trust information (secondhand information) are both used to calculate the trust value. However, some advanced attackers create false recommendation trust information. Therefore, the trust framework needs to distinguish between true and false recommendation information. This paper proposes a novel cooperative recommendation trust model called CRTrust. In CRTrust, the nodes providing recommendations are divided into different sets on the basis of the recommendation similarities. Similar opinions are fused by Dempster's rule and delivered by highly trusted nodes. CRTrust is evaluated both on numerical simulations and SITL(System In The Loop) platform. Simulation results show that CRTrust significantly outperforms conventional trust model without cooperation mechanism.