基于进化优化的并行程序通信边覆盖测试数据自动生成

Software testing is a pivotal mean of guaranteeing software quality, and automatic generation of effective test data is the premise of improving the efficiency of software testing. Since a parallel program is significantly characterized by the communication among multiple processes, sufficient communication testing is a crucial approach of guaranteeing reliabilities of parallel programs. This project studies the issue of automatic generation of test data for communication edge coverage based on evolutionary optimization and expects to obtain the following achievements: (1) proposing the theory and method of selecting target communication edges based on executive relation; (2) constructing the mathematical model for the problem of test data generation of covering the communication edges; (3) presenting the method of automatically generating test data for communication edge coverage based on evolutionary optimization; (4) applying the proposed theory and method in the testing of realistic parallel programs. The achievements of this project can enrich the theory of software testing, improve the efficiency of parallel program testing, and extend the scope of applying evolutionary optimization methods. This project belongs to an intersection of such disciplines such as computer, mathematics and automation, and is explicitly demanded by industry. Therefore, it is very important significance in theory and application.

软件测试是保证软件质量的关键手段，而自动生成有效的测试数据是提高软件测试效率的前提。鉴于多个并行执行进程之间的通信是并行程序的核心特征，因此，充分的通信测试是保证并行程序可靠性的重要途径。本项目研究并行程序通信边覆盖测试数据自动生成理论与应用。通过研究，拟提出基于执行关系的目标通信边选择理论，建立基于分支转化的通信边覆盖测试数据生成问题的数学模型，给出基于进化优化的测试数据自动生成方法，并将所提理论与方法应用于实际的并行程序测试中。研究成果将丰富软件测试理论，提高并行程序测试效率，拓展进化优化方法的应用范围。本项目属于计算机、数学和自动化等学科的交叉，有明确的产业需求，具有重要的理论意义和实用价值。

软件测试是保证软件质量的关键手段，而自动生成有效的测试数据是提高软件测试效率的前提。鉴于多个并行执行进程之间的通信是并行程序的核心特征，因此，充分的通信测试是保证并行程序可靠性的重要途径。本项目研究并行程序通信边覆盖测试数据自动生成理论与应用。项目负责人及成员对基于进化优化的通信边覆盖测试数据生成问题进行了为期3年的研究，提出了基于执行关系的通信边选择理论，转化通信边覆盖问题进而建立了通信边覆盖测试数据生成问题的数学模型，给出了基于进化优化的测试数据自动生成方法，并将所提理论与方法应用于实际的并行程序测试中。发表学术论文10篇，授权发明专利2项，培养博士研究生2名，硕士研究生2名，研究成果丰富了软件测试理论，提高了并行程序测试效率，拓展了进化优化方法的应用范围。