Physical signal-driven fusion for V2X misbehavior detection

While source authenticity and message integrity in vehicle-to-everything (V2X) communications can be well protected via cryptographic schemes, the validity of shared data raises tremendous concerns. An insider attacker can use a legitimate identity to pass the message authentication and disseminate false reports of information such as its location to the surrounding receivers. Without a reliable filter, the control of a host vehicle can be easily trapped to unexpected situations such as wrong acceleration, and that may lead to a crash, loss of life and assets. This work presents a novel misbehavior detection mechanism that can detect such insider attacks. Unlike prior work, our framework works based on the cooperative efforts of (1) the physical signal-based target tracking engine on the host vehicle and (2) the extensive RSSI-based verification of the presence of the target vehicles with spatial and temporal resolution from reliable entities such as road-side units (RSUs). The target is to verify the senders' presence at their claimed locations, validate the consistency of the trajectory movement, and further re-evaluate the reliability of their V2X messages, even without the applicability of radar and LIDAR (e.g., in a non-LOS area). We also demonstrate the example of using this work in various attack scenarios.