Software Reliability Model Analysis Including Internal Structure Based on Bayesian Network

With the improvement of quality of software, software reliability becomes more and more important. Until now, there have been many software reliability models, but the few is suitable for all software projects. Moreover, in the application, it is hard to apply them effectively, because of the operational and other factors, this paper full analyzes the shortcomings in the current models of software reliability and uses Bayesian network to build a new software reliability model based on the internal structure of software systems. Compared to other models, the advantages are as follows: first, the proposed model has not only full combination of expert knowledge and experimental data, but also has a clear intuitive expression of the relevant factors affecting the software reliability; second, the model has a solid mathematical foundation that made the calculate result or inference more accurate; third, it has a good operational and a wide range of practical. This model can be applied to test the reliability of independent developed software system, also can applicable to co-operative software system development process; finally, we used a real example to prove the feasible and practicality of our proposed model.