Detection of Anomalies in Electric Vehicle Charging Sessions

Abstract

Electric Vehicle (EV) charging involves a complex system with cyber-physical components, backend systems, and communication protocols. A potential security incident in this system can open up cyber-physical threats and, for instance, lead to EV battery fires or power grid blackouts. In this paper, we propose a hybrid Intrusion Detection System (IDS) method consisting of regression-based charging session forecasting and anomaly detection. The method considers an EV's detailed charging behavior throughout a session and we discuss and evaluate different design choices. For anomaly detection, we consider both classification- and novelty-based models as well as an ensemble method to combine both models. We perform evaluations based on real-world EV charging session data with simulated attacks. Our results show that regression-based forecasting provides a significant increase in detection performance for attacks affecting individual reports during a charging session. Additionally, the proposed ensemble method, which combines artificial neural network-based classification and local outlier factor-based novelty detection, can maintain a low false alarm rate while offering good detection performance w.r.t. known attacks as well as generalization to previously unseen attacks. We thus argue that the proposed solution can provide a positive contribution to EV charging security, resilience, and trustworthiness.