A CAN Bus anomaly Detection Based on FLXGBoost Algorithm

Along with the networking and automation of smart vehicles and the accelerated application of information and communication technologies, Controller Area Network (CAN) has become the most commonly used protocol in automotive bus networks, but it cannot protect the in-vehicle network from attacks because it does not provide a security system against network and physical attacks, so it is necessary to develop anomaly detection countermeasures to secure the CAN bus. This article proposes a CAN bus anomaly detection model for masquerade attacks, mining the distribution characteristics of each byte of Data messages of all IDs without the bus message identification function, combining the information entropy of message IDs to compose a new data set; introducing a focal loss function to improve the XGBoost algorithm for the case of sample imbalance, and realizing the FLXGBoost-based algorithm. The experimental results show that the anomaly detection model has a high accuracy rate in resisting tampered data domain messages.