摘要： 探讨了建立基于路径使用的Markov链模型来分析模块内部代码结构的统计测试方法的可能性。提出将统计理论用于软件的结构测试，从理论上得出软件的结构可靠性结论。提出具有有限状态、离散时间序列的Markov链作为程序的使用模型和测试模型，使用Kullback判别式作为测试链收敛到使用链的判断准则，从理论上证明了测试链到使用链收敛的必然性。理论分析和初步的实验证明该方法是可行和有前途的。

关键词: 软件统计测试, 结构测试, Markov链, 路径使用模型, Kullback判别式

Abstract: A Markov chain model based on path usage model is developed to analyze the inner code structure in statistical software test. The statistical theory is used on software structure test to obtain the reliability of the software structure theoretically. A Markov chain with finite state, scattered time sequence is put forward as the usage model and test model of the program. The Kullback discriminant is used as the convergence judgment criteria of the test-chain to usage-chain, and the inevitability of the convergence is testified theoretically. The primary experiments and theorized analysis prove the method is approving and promising.

Key words: Software statistical test, Structure test, Markov chain, Path usage model, Kullback discriminant

中图分类号: 

• TP311