Attacker-induced traffic flow instability in a stream of automated vehicles

Automated driving technologies such as Adaptive Cruise Control (ACC) are becoming a standard package on mid-grade vehicles. An ACC system utilizes local sensors to monitor the motion of surrounding vehicles. Feedback from these sensors is used to calculate automated braking and acceleration requirements to maintain a specified inter-vehicle spacing. Another near term technology, Cooperative Adaptive Cruise Control (CACC), will augment the data collected by the local sensors of ACC with inter-vehicle communication. This will allow vehicles to cooperatively change their state with respect to surrounding vehicles, such as increase their nominal velocity to adjust highway vehicle density to accommodate merging vehicles. As these technologies become mainstream the widely researched concept of Automated Highway Systems (AHS) is quickly becoming a reality.;Extensive research over the last several decades has proposed various ACC and CACC schemes as a solution to highway automation. The stability of these designs has been analyzed in great detail. However, the field of research largely assumes all the vehicles in a traffic system of automated vehicles are operating in a non-adversarial environment. This work highlights the error of such assumptions and considers attacks against the system where some members act in a malicious manner with the intent of destabilizing traffic flow. This type of attack could consist of active attackers who modify their control gains to affect neighboring vehicles, or passive attackers whose gains have been changed, perhaps without the knowledge of the vehicles driver, and could be remotely activated.;A variety of proposed control algorithms are surveyed as part of this work, several of which have been simulated under attack as an automated traffic system of n-vehicles is considered. The results demonstrate that an attacker can force surrounding vehicles into a never ending cycle of oscillatory braking and acceleration. By placing themselves strategically in a system of traffic colluding attackers can destabilize the entire system. The same can be accomplished by remotely activating the maliciously modified control laws of the unwitting passive attack participants.