软件测试中的若干组合结构研究

With the rapid development of computer technology, software systems become increasingly large. Component-based software development poses many challenges for the software testing. Interactions among components are complex and numerous. Components are prone to unexpected interaction faults. Ideally, one would test all possible interactions, but this is usually infeasible in practice. Under this circumstance, as an effective and scientific approach of software tests, combinatorial testing methods have attracted greatly many researchers from software engineering and combinatorial community, and become a very active research area currently. The research focus is the generation of software test suites，that is, how to build the smallest possible test suite in order to minimize test cost and improve test efficiency under the premise of ensuring the capability of detecting/locating interaction-faults for the given software system to be tested. This project aims to the present challenges software tests face and the forefront of combinatorial designs. It will investigate the existence, construction, optimality and related problems for several combinatorial configurations being test suits, which include variable-strength covering arrays, locating arrays and detecting arrays et al. Research results of this project will provide new combinatorial methods and new test suites for software testing.

随着计算机技术的飞速发展，软件系统变得越来越庞大。软件开发带来了对于软件系统测试的许多挑战。组件之间的交互作用量大而且复杂，难以预料。在理论上，人们可对组件之间所有可能的交互作用逐一进行测试， 但这样的穷试方法在实践上显然是不可行的。在这样的背景下，组合测试作为一种科学、有效的软件测试方法受到了组合数学和软件工程领域中诸多学者的高度重视和密切关注，成为目前非常活跃的一个研究领域。研究的焦点是组合测试用例集的生成问题，即如何针对具体待测软件系统，构建规模尽可能小的测试用例集,以便在保证错误检测和定位能力的前提下尽可能降低测试成本、提高测试效率。本项目瞄准当前软件系统测试面临的挑战和组合设计领域前沿，研究作为软件测试用例集的变强度覆盖阵、定位阵和检测阵等若干组合结构的构造问题、优化问题及其相关问题。项目研究成果的取得将为软件测试提供新的组合方法和新的组合测试用例集。

随着计算机技术的飞速发展，软件系统变得越来越庞大。软件开发带来了对于软件系统测试的许多挑战。组件之间的交互作用量大而且复杂，难以预料。在这样的背景下，组合测试作为一种科学、有效的软件测试方法受到了组合数学和软件工程领域中学者的高度重视和密切关注。本项目对软件测试用例集包括覆盖阵、定位阵和检测阵的构造问题和最优性问题，以及相关的问题展开了较为深入的研究，取得了实质性的研究进展。 研究成果包括：给出了构造任意强度正交阵列的叠加方法，从而解决了Hedayat, Sloane and Stufken 提出的一个问题 （Orthogonal Arrays, pp. 220）；找到了混合水平检测阵列的最优性判别准则， 并运用组合方法和优化算法构造了多类达到该下界的最优检测阵列，为软件测试提供了新的组合方法和新的组合测试用例集。项目实施的四年间，项目组在SCI源期刊上发表学术论文14篇，其中 1 篇发表在统计学顶级期刊《The Annals of Statistics》上，4 篇发表在信息论顶级期刊《IEEE Trans. IT》上。