A Trust System for Detecting Selective Forwarding Attacks in VANETs

Abstract

Vehicular ad-hoc Networks (VANETs) have inherent high mobility and take data forwarding as a basic mechanism to share information among vehicles. Selective forwarding attack, in which malicious nodes deliberately drop data packets, destroys the integrity of data and hurts the validity of real VANETs applications. Because malicious nodes usually masquerade themselves as normal nodes and collude with each other whenever possible, it is hard to obtain clear and direct evidence of selective forwarding attacks. In this paper, we address the issue of detecting selective forwarding attacks by building a trust system. The proposed approach to maintain this system mainly includes (1) local and global detection of attacks based on mutual monitoring among all nodes, and (2) detection of abnormal driving patterns of malicious nodes. Since both in-band and out-band information is utilized, our approach is effective in relatively low-density road conditions and resilient to various scenarios, such as different rate of malicious occurrence or different road’s range. The extensive simulations demonstrate that our approach achieves a high fault tolerance by choosing most reliable nodes for information delivery, while at the same time identify malicious nodes with relatively high accuracy.