Design and Verification of Distributed Recovery Blocks with CSP

A case study on the application of Communicating Sequential Processes (CSP) to the design and veriication of fault-tolerant real-time systems is presented. The distributed recovery block (DRB) scheme supports the uniform treatment of hardware and software faults in real-time computer systems. DRB combines distributed computing with the recovery block scheme which is well known for handling software faults. Instead of modelling hardware faults explicity as events, faulty components are modelled as nondeterministic processes in the case study. Safety and liveness properties of the DRB design are speciied and veriied using the untimed version of CSP whereas real-time requirements are treated using the timed version of CSP, which is compatible with the untimed version. The rich set of CSP operators and their assoicated algebraic rules, together with the well-structured proof systems, are found to be well suited to the design and veriication of DRB.