First quantitative results of the dependability improvement achieved by ReCANcentrate

Abstract

There is a growing interest in using star topologies instead of buses as the communication infrastructure for highly-reliable distributed control systems, given the better dependability stars are supposed to provide. For the Controller Area Network (CAN), we developed a simplex and a replicated star called CANcentrate and ReCANcentrate respectively. In a previous work we modelled the dependability of the CAN bus and CANcentrate using Stochastic Activity Networks (SANs). There we presented the first quantitative analysis of the error-containment benefits of a simplex star when considering permanent hardware faults. This paper quantitatively analyzes, for the first time, how a replicated star such as ReCANcentrate can improve both error-containment and reliability, also considering permanent hardware faults. We explain our modelling strategy using SANs and show some first and novel results.