Tree-Based Intermittent Connection Fault Diagnosis for Controller Area Network

Controller area network (CAN) is the most widely adopted network protocol in vehicle communication systems. The intermittent connection (IC) fault, a random cable connectivity problem with short durations, occurring in the CAN network may result in system performance degradation and system level failures in severe cases. Therefore, the health management of CAN-based networked systems demands effective detection and accurate localization of the IC fault. However, the diagnosis of IC fault in CAN network has not been fully exploited. This paper proposes a novel tree-based IC fault diagnosis method to identify the various IC fault locations in the CAN network. First, the error event pairs for network nodes are defined by analyzing the data-link layer information upon each error. Then, the domains of error event pairs and the information cost of IC faults included in the domains are derived. Finally, a tree-based minimal cost diagnosis algorithm is proposed to find the locations of the IC faults. Laboratory testbed is constructed and the case studies are conducted to demonstrate the performance of the proposed method. The results show that IC fault identified by the proposed method agree well with the case study setups.