Detecting and Mitigating Spoofing Attack Against an Automotive Radar

Applying cyber security mechanisms, such as cryptography, trusted computing, and network intrusion detection, is insufficient to secure cyber-physical systems (CPS) such as smart vehicles, because most sensors (and actuators) are designed without security considerations and remain vulnerable to sensor spoofing attacks in the analog domain. To address this problem, we present a new approach to detect and handle sensor spoofing attack against automotive radars-a key component for assisted and autonomous driving-by extending multiple beamforming in an automotive multi-input multi-output (MIMO) radar. In a simulation study for adaptive cruise control based on the car following model, our approach significantly outperforms three state-of-the-art baselines in terms of attack detection and ranging accuracy.