Detection and Mitigation of Data Integrity Attacks in Vehicle Platoons Using Unknown-Input Kalman Filters and Penalized Weighted Least Squares.

This study enhances the security of vehicle platoon systems against false data injection attacks (FDIAs), which can compromise critical states, such as position, velocity, and acceleration. The proposed approach ensures safe and stable autonomous vehicle coordination by detecting and mitigating such attacks under flexible platoon conditions. Unlike the existing methods, it supports arbitrary communication topologies, varying platoon sizes, dynamic lead-vehicle behavior, and accounts for process and measurement noise-making it applicable to a wide range of real-world scenarios. A state-space model is formulated to capture the system dynamics, and an unknown input Kalman filter (UIKF) is employed to estimate both vehicle states and the lead vehicles acceleration as an unknown input. To counteract malicious data, a penalized weighted least squares (PWLSs) method is used to downweight suspicious measurements based on discrepancies between the observed and estimated values. The simulation results demonstrate the effectiveness of the proposed framework in detecting and mitigating FDIAs.