软件的单故障模型及故障的测试方法

研究表明,软件中残留的故障一般都是难以测试,这些故障往往都是局部的、检测概率比较“小”的故障,也就是说,和正确的软件相比,这些故障的表现往往只是一个或几个符号发生错误,现提出的软件单故障模型正是基于这个原理。测试方法是以面向故障为基础,将故障的检测分为3个部分:故障语句的执行,故障的激活和故障的传播。为加速测试用例的生成,也同时提出了软件的故障模拟方法。最后,给出了部分实验结果。

Software's Single Fault Model And Its Testing Methods

It is demonstrated that the remaining faults of software are very difficult to detect. Generally such faults are partial defects with smaller detecting probability. Comparing with correct software, only one or several symbols are incorrect, such as operator faults, variable faults, incorrect parentheses, etc. The single fault model of software is based on the principle. The fault detecting methods are designed to deal with single fault model, which can be divided into three parts: executing the sentence including fault, activating fault and propagating fault. The fault simulation methods are proposed to accelerate the generation of test cases. The basic experimental results demonstrate that the single fault model of software and its fault testing methods are efficient in finding software faults.