摘要
提出了一个改进的马尔科夫决策过程的软件测试模型,应用交叉熵方法计算求解改进后的测试模型下的软件测试优化策略,得到最优测试剖面,使得平均测试费用最小.并对采用随机软件测试策略,原始的MDP模型软件测试策略和改进后的MDP模型软件测试策略的软件测试过程进了仿真.仿真结果表明,改进后的软件测试策略不仅能够大大降低期望测试费用,而且也减少了测试用例的使用数量,提高了软件测试的效率和有效性.
This paper gives an improved model for software testing to obtain optimal testing profile with lower cost and studies the computation of the improved model based on cross-entropy. We simulate the software testing process by using random software testing strategy, strategy of the original MDP model and the improved MDP model. Simulation results show that the improved MDP software testing model strategy is not only greatly reduce the expected cost of testing, but also requires the less number of test cases which improves the efficiency and effectiveness of software testing.
出处
《数值计算与计算机应用》
CSCD
2014年第2期92-102,共11页
Journal on Numerical Methods and Computer Applications
基金
国家部委基础研究资助项目(A2120110006)
关键词
软件测试
最优测试剖面
交叉熵
马尔科夫决策过程
software testing
optimal testing profile
cross-entropy
Markov decision process
