摘要
基于统计测试的马尔可夫使用模型对软件可靠性评估提出了一种有效的估计方法.该方法利用重要抽样技术在保证可靠性估计无偏性的条件下,利用交叉熵度量操作剖面与零方差抽样分布之间的差异,通过启发式迭代过程调整各个状态之间的转移概率来修正测试剖面.从理论上证明了利用修正测试剖面测试估计的可靠性是方差为0的无偏估计.最后给出了软件可靠性估计的最优测试剖面生成的启发式迭代算法.仿真结果表明,该方法与模拟退火算法相比,能够明显降低估计的方差,在提高估计精度的同时加快统计测试速度.
This paper proposes an effective method for computing optimal state transition probabilities for software reliability estimation based on a Markov usage model. This method uses Cross-Entropy to measure the differences between the operational profile and the sampling distribution with zero variance. By adjusting the probabilities of state transitions during test, an iterative method based on the Cross-Entropy is proposed for this choice, and an unbiased reliability estimator with zero variance is obtained. Simulation results show that the testing profile with Cross-Entropy method performs significantly better than the simulated annealing algorithm. Moreover, this strategy can more effectively accelerate software statistical testing.
出处
《软件学报》
EI
CSCD
北大核心
2009年第10期2859-2866,共8页
Journal of Software
基金
国家自然科学基金Nos.60773104
60721002
90818027
国家高技术研究发展计划(863)Nos.2008AA01Z143
2009AA01Z147~~
关键词
软件可靠性
统计测试
马尔可夫使用模型
重要抽样
交叉熵方法
software reliability
statistical testing
Markov usage model
importance sampling
cross-entropy method
