Lightweight Kalman Spoofing Detection in Platoons of Vehicles

Spoofing attacks remain among the most critical security threats in Connected and Autonomous Vehicles (CAVs). This work introduces a lightweight, two-level spoofing detection framework based on Kalman filtering, designed for real-time deployment in vehicular platoons that communicate over Dedicated Short-Range Communications (DSRC). At the first level, a heuristic residual-based detector identifies abnormal measurement deviations using adaptive statistical thresholds. At the second level, a Mahalanobis distance test assesses model consistency using covariance-aware anomaly scoring at a 95% confidence level. The combination of these complementary mechanisms enables both rapid alerting and robust statistical verification without the need for machine-learning training or centralized processing. Simulation results from 20 independent nodes demonstrate that the proposed approach achieves an average F1-score of 0.92 and Area Under the ROC Curve (AUC) of 0.72, outperforming standalone detectors while maintaining low computational cost. Compared with deep learning and adaptive Extended Kalman Filter (EKF) approaches, the proposed framework achieves similar detection performance while substantially reducing computational complexity and enabling full real-time operation, making it suitable for embedded in-vehicle security modules.