Single-use reliability computation of a semi-Markovian system. (Q489240)

The author proposes arc-based semi-Markov usage models to test systems, extending the prominent approaches of \textit{J. A. Whittaker} and \textit{M. G. Thomason} [``A Markov chain model for statistical software testing'', IEEE Transactions on Software Engineering 20, No. 10, 812--824 (1994; \url{doi:10.1109/32.328991})] on failure state models and the arc-based Bayesian models of \textit{K. D. Sayre} and \textit{J. H. Poore} [``Stopping criteria for statistical testing'', Information and Software Technology 42, No. 12, 851--857 (2000; \url{doi:10.1016/s0950-5849(00)00110-5})]. In particular, the author extends previous studies that rely on the Markov chain assumption to the more general semi-Markovian setting. Moreover, the author analytically computes the expectation and variance of the system's single-use reliability, with the use of the geometric transform, thus extending the previous Markov chain approach. The suggested model allows for an increased flexibility through the consideration of reliabilities that depend on the states of the system and on the length of stay in the initial state. This implies that the transition reliability depends not only on the specific arc (system instruction) to be executed, but also on the time of execution that is, in general, random.