Reliability improvement achievable in CAN-based systems by means of the ReCANcentrate replicated star topology

Despite there is a growing interest in using star-based networks for distributed control systems, the reliability benefits of this topology have not been appropriately quantified. Thus, in previous work we modelled, by means of Stochastic Activity Networks (SANs), the system reliability that can be achieved with CAN and a simplex star we proposed for it called CANcentrate. These models quantitatively compared the system reliability of CAN and CANcentrate when permanent hardware faults occur. However, the strategy that underlies these models was not suitable, in terms of computation time, for ReCANcentrate, a replicated star we more recently proposed for CAN. This paper thoroughly describes a new model that overcomes this drawback and, then, demonstrates that ReCANcentrate can boost the system reliability. Although this paper addresses CAN-based networks, the results herein presented can be extrapolated to other field-bus technologies.