Software reliability modeling and analysis

Abstract

This thesis investigates the modeling problem of software reliability, extendingtraditional reliability models through relaxing some specific restrictive assumptions.Related analysis issues, especially resource optimization and early predictionproblems, are addressed with the corresponding extended models. Centered on thisline, research has been developed through the following related sub-topics. Firstly,extended software reliability modeling approaches are proposed through combiningboth FDP (fault detection process) and FCP (fault correction process). Secondly, earlyreliability prediction problem is also studied with both analytical and data-drivenmodels. Comprehensive case studies are carried out leading to some interestingempirical conclusions. Thirdly, study has also been conducted to model the softwarereliability with Markov Bayesian networks. Also, the corresponding method to solvethe network model is developed. At last, for the Bayesian networks application insoftware reliability, we also explore the issue of software inspection effectivenessanalysis. Finally, some related topics for further research have been discussed forcontinuing research in the future.