Dependability evaluation of fault tolerant architectures in distributed industrial control systems

Abstract

Modern distributed industrial control systems need improvements in their dependability. We study different fault tolerant architectures for the nodes of these systems and present three different alternatives in order to develop fault tolerant nodes. Moreover, in order to evaluate their dependability we present theoretical models of each one, based on Markov chains, and the results obtained (reliability and safety). With the results obtained we can choose the most appropriate node architecture for one particular distributed industrial control system as a function of the reliability and safety constraints, and the complexity and cost of each alternative.