Central misbehavior evaluation for VANETs based on mobility data plausibility

Trustworthy communication in vehicular ad-hoc networks is essential to provide functional and reliable traffic safety and efficiency applications. A Sybil attacker that is simulating "ghost vehicles" on the road, by sending messages with faked position statements, must be detected and excluded permanently from the network. Based on misbehavior detection systems, running on vehicles and roadside units, a central evaluation scheme is proposed that aims to identify and exclude attackers from the network. The proposed algorithms of the central scheme are using trust and reputation information provided in misbehavior reports in order to guarantee long-term functionality of the network. A main aspect, the scalability, is given as misbehavior reports are created only if an incident is detected in the VANET. Therefore, the load of the proposed central system is not related to the total number of network nodes. A simulation study is conducted to show the effective and reliable detection of attacker nodes, assuming a majority of benign misbehavior reporters. Extensive simulations show that a few benign nodes (at least three witnesses) are enough to significantly decrease the fake node reputation and thus identify the cause of misbehavior. In case of colluding attackers, simulations show that if 37% of neighbor nodes cooperate, then an attack could be obfuscated