TECS: A Trust Model for VANETs Using Eigenvector Centrality and Social Metrics

Vehicular Ad Hoc Networks (VANETs) rely heavily on trustworthy message exchanges between vehicles to enhance traffic efficiency and transport safety. Although cryptography-based methods are capable of alleviating threats from unauthenticated attackers, they can not prevent attacks from those legitimate network participants. This paper proposes a trust model to deal with attackers from the latter case, who can tamper with their received messages and deliberately decrease the trust value of benign vehicles. The trust evaluation process is formed by two stages: (i) the local trust evaluation at vehicles and (ii) trust aggregation on Road Side Units (RSUs). In the local trust evaluation stage, vehicles detect attacks and calculate the trust value for others in a distributed manner. Also, the social metrics of vehicles are calculated based on interaction records and trajectories. In the trust aggregation stage, each RSU collects local data from nearby vehicles and derives aggregation weights from the eigenvector centrality of the local trust network and social metrics. Then the RSU broadcasts the aggregated trust value towards vehicles in proximity. These vehicles can thus obtain a more accurate and comprehensive view. Vehicles with trust value below a preset threshold will be considered malicious. Extensive simulations based on the ONE simulator show that the proposed model (TECS) outperforms another benchmark model (IWOT-V) regarding the malicious vehicle detection and the delivery rate of authentic messages.