Multi-Layer Edge Intrusion Detection for Power Electronic Systems Using CAN Bus and Wireless Traffic Monitoring in Maritime Applications
Abstract
Electric maritime vessels rely on interconnected power electronic systems for propulsion, energy and battery management, and auxiliary services. These systems typically communicate via Controller Area Network (CAN) buses while also incorporating wireless connectivity for remote monitoring and cloud services. Although such connectivity enhances operational efficiency, it introduces cybersecurity vulnerabilities that can compromise the reliability of critical maritime infrastructures. In this context, this paper proposes an edge-based cyber-physical monitoring framework to enhance the resilience of maritime power electronic systems by jointly supervising operational data and wireless communications. The architecture consists of two monitoring layers deployed on a resource-constrained edge device. The first layer performs real-time CAN monitoring, decoding signals through a CAN database (DBC) to extract electrical parameters such as voltage, current, and temperature, and to detect abnormal operating conditions. The second layer implements a machine learning–based intrusion detection system using a Random Forest classifier to analyze wireless network traffic and identify potential cyberattacks. The monitored data and detected anomalies are transmitted to a cloud platform for centralized logging and remote visualization. Experimental results demonstrate successful detection of CAN operational anomalies and intrusion detection accuracy exceeding 99% for wireless attacks, while maintaining low computational overhead.