A software safety test approach based on FTA and Bayesian networks

As an important way to verify software safety, software safety test has caught more attentions in practice. However, it is still an open question that how engineers could make software safety test more efficient. Currently, FTA based method is one of the approaches in software safety test, but it can not utilize the finished software test results, and can not be determined the priorities of all the use cases. In order to solve these problems, this paper gives a quantitative approach of software safety test based on FTA and Bayesian networks. In the approach, top-level events of fault trees are identified from system hazards firstly. Then, fault trees are built using FTA and transferred into Bayesian networks. Finally, test cases of software safety test are determined by the Bayesian networks. Besides, the paper also shows an example using the approach, which could guide software engineers to make software safety test more efficient. The example shows that the approach could take advantage of Bayesian Theorem and FTA methodology together, and give reasonable priorities of use cases in software safety test.