基于测试用例的软件可靠性分析

The software reliability analysis based on test cases

软件可靠性问题是软件测试中的重要难题之一,由于程序语言的复杂性以及被测程序的多样性,一直没有一个好的可靠性模型来评价它。鉴于此,提出了一个有关黑盒测试的可靠性分析模型,模型在结合等价类和边界值划分的基础上充分考虑软件复杂性、缺陷后果严重性和信息不足等因素等原因,根据矩阵和数理统计的知识得出相关的可靠性模型,并使用EM算法对该模型的置信下限做出了相应的理论证明。实验表明,该模型在实用性和可靠性方面都取得了较大的进展。

Software reliability is one of the important difficult problems in software test. Because of the complexity of the programming languages and the variety of applications, there is not a good module to evaluate the software reliability. Therefore, a new module was presented based on black box test, which incorporated equivalent class and boundary value to analyze software complexity, fault severity and information insufficiency by matrix and mathematical statistic knowledge, and proved confidence lower limit of this model by EM algorithm. Finally, the result of experiment shows this model has a better performance.