In-vehicle CAN Bus Anomaly Detection Algorithm Based on Linear Chain Condition Random Field

Abstract

Aiming at solving of missing and false alarm in anomaly detection of In-vehicle CAN communication, which only detects a single CAN message but ignores the correlation among multiple messages, an anomaly detection algorithm of vehicular CAN bus based on linear chain conditional random field (Linear-CRF) is proposed in this paper. Based on L-BFGS (Limited-memory BFGS) learning algorithm, the double-contour modes of Linear-CRF normal model and anomaly model are constructed respectively. The algorithm of message anomaly detection based on Linear-CRF model is elaborated in detail. The proposed anomaly detection algorithm is simulated and analyzed. The simulation results show that the proposed algorithm can effectively identify the relevance of the context content of the message, and effectively improve the detection accuracy of the In-vehicle CAN bus anomaly message.