Reliability modeling of network fabric fault-tolerant industrial communication systems

Without fault-tolerance, a single component failure in a Networked Control System (NCS) can cause system downtime and potentially large production losses. Thus, fault-tolerance is fast becoming a key aspect in the design and evaluation of NCS architectures. In this paper, the focus is on the reliability modeling of network fabric level fault-tolerant NCSs. The two main considered fault-tolerant architectures are based on the Parallel Redundancy Protocol (PRP) and the Rapid Spanning Tree Protocol (RSTP) respectively. Additionally, reliability modeling of a simplex architecture, where no fault-tolerance is implemented, is considered as a baseline for comparison of both fault-tolerant architectures. Moreover, a case study is presented to compare the reliability of the fault-tolerant architectures under study using typical industrial parameters. It is shown that the RSTP architecture offers higher system reliability and longer mission time compared to the other studied architectures.