Software reliability assessment based on Bayesian method

Since lack of reliability metrics in software testing process, a new software reliability assessment method was proposed based on Bayesian theory. An analysis was made of the relationship between software modules and Bayesian networks was established, and then the empirical data were adopted to assess the reliability of each node. The detailed steps to calculate numerical reliability of software modules were introduced and the calculation process was illustrated by a specific example. With the method, the existing test results could be resorted to and the calculation became faster. Numerical reliability assessment for each module provided the basis of test case selection and the test cost reduced, as a result, a good support for software testing was offered.