Intermittent Connection Fault Diagnosis for CAN Using Data Link Layer Information

Controller area networks (CANs) had been applied to many safety critical applications, hence the reliability of the CAN network becomes increasingly important. Due to the aging of the cable connections and inappropriate maintenance, the intermittent connection (IC) problem of network cables will occur on the CAN network over time, which will affect the system performance, deteriorate the reliability and safety of the system. Hence, effective detection and accurate localization of the IC fault are crucial to the system reliability assurance. In this paper, a novel IC fault localization algorithm for CAN-based systems is proposed, which utilizes the data link layer information of the CAN network for IC fault pattern analysis. First, the fault events are defined by comparing the detected error records. Second, the context free grammar is used to represent the IC faults based on network topology information. Then, based on the node error event pattern correlation, the concurrent localization algorithm is proposed to localize the local and backbone IC faults. Testbed is constructed and case studies are conducted to demonstrate the proposed methodology under various fault scenarios. Experiment results show that the IC fault locations identified by the proposed method agree well with the experiment setup.