一种基于贝叶斯理论的软件可靠度评估方法

Software Reliability Assessment Based on Bayesian Method

摘要:

针对目前软件测试过程中可靠性度量方法匮乏的不足，提出了一种新的基于贝叶斯理论的软件可靠度评估方法，该方法通过分析软件模块间的影响关系构建贝叶斯网络，利用贝叶斯方法和经验数据评估贝叶斯网络中各节点的可靠度。给出了可靠度计算的详细推导过程，并通过一个具体实例解释了计算过程。该方法能重用已有的测试结果，计算速度快，简单易行。通过对每个模块提供数值可靠度估算，可以作为测试用例选择的依据，降低了测试成本，从而很好地为软件测试提供支持。

Abstract:

Since lack of reliability metrics in software testing process, we presented a new software reliability assessment method based on Bayesian theory. In the method, we analyzed the relationships between software modules and constructing Bayesian networks, and then used the empirical data to assess the reliability of each node. The paper described the detailed steps of calculating numerical reliability of software modules, and explained the calculation process by a specific example. This method can reuse the existing test results and make the calculation faster. Numerical reliability assessment for each module can be the basis of test case selection and reduced the cost, thus providing good support for software testing.