Limits on Revocation in VANETs

We examine the limitations on revocation approaches in VANETs, including local revocation and global revocation. Local revocation schemes often use a local vote to identify and revoke an attacker. However, such votes often require that not only a majority of local nodes are honest, but that they also are able to detect the attack. We argue that these requirements may not be practical, particularly in the early stage of VANET deployment. Another local revocation approach, RevoGame [1], uses game theory to mitigate misbehavior in the VANET; however, we argue RevoGame does not correctly identify the players in the revocation game. We also analyze the limits of global revocation based on the misbehvaior accusations made and evidence gathered by each vehicle. Our analysis shows that no algorithm that uses only the accusation graph can identify attackers without false positives and false negatives.