In: IEEE Transactions on Reliability. Special Issue on Fault Tolerant Software, Vol.42, N°2, pages 205-219. June 1993.
Abstract: Fault-tolerant software approaches have given rise to numerous reliability models. However, all these models assume stable reliability, i.e., they do not consider reliability growth resulting from progressive removal of design-induced faults. This paper is aimed at modeling and estimating the influence of reliability growth of the fault-tolerant software components on the reliability of the overall software in operational life. Two fault-tolerant software techniques are investigated: recovery block and N-version programming. For these two techniques, the stable reliability models are transformed into models that take account of reliability growth via the transformation approach based on the hyperexponential model. Analytical and numerical processing of the transformed models enable the influence of fault removal on the reliability of the considered fault-tolerant software approaches to be identified.
Keywords: fault tolerant software; reliability growth; Generalized Stochastic Petri Nets; GSPN.