Robust Cybersecurity for Autonomous Vehicles Using Particle Filter Based Anomaly Detection

This paper addresses the critical challenge of detecting and interpreting cybersecurity anomalies in Autonomous Vehicles (AVs) under high-frequency cyberattacks using a Particle filter. In this approach, we leverage the power of Particle filter-based state estimation, combining it with suitably defined thresholds and anomaly detection metrics to detect cyberattacks. In addition, to demonstrate the superior performance of the Particle filter for cyberattack detection, a comparison between the Kalman filter and the Particle filter has been conducted. The simulation results performed on the HuskyA200 autonomous ground vehicle (AGV) demonstrated that the Particle filter provides superior performance and interpretability during high-frequency attacks compared to the Kalman filter. The feedback from Particle filter-based detection can help the control functions of the vehicle, such as velocity damping and orientation correction, mitigate attack impacts for real-time operation.