软件可靠性评测的多因素决策模型

Multi-factor decision making for software reliability evaluation

提出了一种利用软件生命周期过程中影响软件可靠性的软件质量特性进行软件可靠性评测的方法。从软件失效的机理出发,分析并提取了刻画软件可靠性各个方面的因素,同时借助模糊分析及专家系统的理论对软件可靠性因素进行了定量的划分和描述。针对软件可靠性评测的多因素决策问题,提出了一种基于随机森林的软件可靠性评测模型。通过蒙特卡罗模拟仿真建立了各个可靠性因素的概率模型以获得评测样本数据集,并在此数据集上对所提出的评测模型进行了评测和分析。实验结果表明,所提出的方法能够对软件可靠性进行准确的评测,且不依赖于特定的可靠性因素的先验概率,显著提高了软件可靠性的评测性能。同时验证了该方法能够克服小样本集上易出现的过拟合及表现力差的问题,具有较好的稳定性和鲁棒性。

This paper described an accurate and efficient software reliability evaluation approach by using software quality features that affected the software reliability in the whole process of software development lifecycle. It extracted reliability factors to characterize the software reliability from the perspective of software failure,and employed fuzzy analysis and expert system to quantificationally depict those factors. Then to address the problem of multifactor decision of software reliability,it established a novel software reliability evaluation model based on random forest. Monte Carlo simulation was utilized to acquire the distribution of each reliability factor and then further to generate simulation data samples for evaluation. Extensive experimental results on these data samples show that the proposed approach can make accurate predictions and estimations,and substantially improve the evaluation accuracy compared with traditional approaches. Additional experiments verify the effectiveness of this approach at different training sample sizes,showing its stability and robustness on smaller training data set.