An Effective Integrated Reliability Modeling for Safety‐Critical FlexRay Communication Systems

ABSTRACT A primary issue in FlexRay communication system design for safety‐critical automotive applications is how to verify the system reliability in early design phase. Reliability modeling is often used to predict the system reliability before its implementation. Previously, research rarely discusses the effects of network topology and clock synchronization between FlexRay ECUs of a cluster on the system reliability. In this study, we first apply a well‐known probabilistic reasoning method, Bayesian network (BN), to model the reliability of FlexRay systems with various network topologies, and then employ the Markov chain (MC) to model the reliability of clock synchronization process in terms of the number and failure rates of sync ECUs in the cluster. The BN and MC are then combined to form an integrated model termed as BN‐MC for thorough assessment of the FlexRay synchronization and system reliability. Through a FlexRay steer‐by‐wire (SBW) case study, we demonstrate the influence of various ECU failure rates, topologies, and number of sync ECUs on the reliability of FlexRay SBW system. Furthermore, based on BN‐MC model we propose a safety‐level‐compliant preliminary system architecture exploration process (SPSAEP) to effectively assist the system engineers in achieving a reliable FlexRay system architecture, which complies with the demanded safety integrity level.