Threat Detection for Collaborative Adaptive Cruise Control in Connected Cars

We study collaborative adaptive cruise control as a representative application for safety services provided by autonomous cars. We provide a detailed analysis of attacks that can be conducted by a motivated attacker targeting the collaborative adaptive cruise control algorithm, by influencing the acceleration reported by another car, or the local LIDAR and RADAR sensors. The attacks have a strong impact on passenger comfort, efficiency and safety, with two of such attacks being able to cause crashes. We also present two detection methods rooted in physical-based constraints and machine learning algorithms. We show the effectiveness of these solutions through simulations and discuss their limitations.