基于可信基因的网格测试分级约简研究

作为保证网格服务质量的重要技术手段，网格测试已成为网格研究的热点内容之一．而软件测试的核心与关键则是测试用例的生成问题，测试用例的数量和质量将直接决定软件测试的成本和有效性．就目前而言，随着覆盖程度的提高，所需使用的测试用例数量也在呈指数级增长，且冗余随之而来．因此，若进行完全测试，势必工作量很大，但若只按一般组合设计测试用例，则又会存在因测试次数太少而牺牲相应检错能力的问题．为此，提出了基于可信基因的网格测试约简方案，并通过实践验证了所得到的测试集不仅规模较小、保持了极佳的检错能力，而且与传统算法相比较时空间复杂度有所下降．

Research on grid test suite reduction based on trusted gene

As an important approach to assure the quality of grid services, the grid software test has been the highlight of grid technology. The core and key of software testing is test-case creating because the quantity and quality of test case determine the cost and validity of software testing immediately. Now, along with improvement of logic covering degree, the quantity of test-case presents exponential growth, and then, followed by redundancy. Because complete test has heavy workload and general creating scheme of test-ease using combination design will probably decrease error-detecting ability, the grid test suite reduction scheme based on trusted gene is presented. Finally, a series of simulative experiments are completed to evaluate the performance of the scheme. The result shows that the scheme is effective.