Reliability Bayesian Verification Scheme for Discrete Software
Yuzhuo Wang, Liu Haitao, Haojie Yuan, Zhai Yali, Zhihua Zhang · 2024
The prior distribution determination methods of existing Bayesian schemes are relatively ideal and subjective in processing reliability growth testing information of discrete software. In this article, a software success probability estimation and its effectiveness test are proposed using the discrete software reliability growth model and information entropy principle. On this basis, a prior distribution determination method is proposed, and a Bayesian verification scheme based on maximum a posteriori risk is developed to protect the interests of users. Two case studies have shown that the proposed prior distribution determination method is more reasonable for processing reliability growth testing information of discrete software. The formulated Bayesian scheme can significantly reduce the number of verification test cases while ensuring the credibility of the scheme, thus having certain engineering application value.