CAN node reliability assessment using segmented discrete time Markov chains

In manufacturing systems, Controller Area Network (CAN) based systems has been widely used. Hence the reliability of the Controller Area Network has become increasingly important. In order to ensure production safety and reduce unnecessary maintenance costs, evaluating the reliability of a CAN network becomes necessary. The reliability of the CAN network can be assessed by determining the status of nodes and predicting the times of each node reaching its bus off states using node transmit error counter (TEC) values. However, the TEC values of the nodes cannot be accessible online in most industrial applications. This paper presents a novel method for node TEC value estimation based on the segmented discrete time Markov chains using network error information. We also consider the special scenario where data frame arbitration area is damaged by the error so that node identifier (ID) information of the interrupted data frame cannot be identified. Case studies are carried out to demonstrate the effectiveness of proposed method. Experiment results show that the estimation results using the proposed method agree well with actual observations.