并行、多平台“云测试”的关键技术研究

With the dramatic increasing of software scales and the complexity of software functionalities, software testing is under tremendous pressure from practical challenges. It is straightforward to parallelize the process of regression testing, but such a brute force solution is infeasible because of the stringent resource constraints. With the rapid developing of mobile devices and networking technologies, cross-platform software shows an observable growing trend, while the implementation of multi-platform software testing is technically difficult. Cloud-based testing service gives a solution to parallel, multi-platform software testing. However, there are many practical issues, such as the combination explosion problem in the multi-platform scenario, the efficiency of exercising large number of test cases, and the simultaneous localization of multiple software faults. In this project, we will start with realistic engineering problems and focus on the key technologies in parallel, multi-platform cloud-based testing. They include combinatorial testing, test case prioritization, and automatic software fault localization. In these active research fields, recent works seldom focus on software engineering issues in the cloud testing scenario. In this project, based on modern software features and software testing problems, we study the key technologies of parallel, multi-platform cloud-based testing. In this project, we will investigate the theories and techniques about how to increase the speed of regression testing, how to improve the quality of continuous integration, and how to enhance the effectiveness of fault localization. We will provide novel ideas and solutions to cloud-based automatic testing and facilitate the implementation of parallel, multi-platform testing for software manufacturers. In this project, we will also make use of combinatorial testing, regression testing, and fault localization techniques to build cloud-based software testing services and toolsets, to ease modern software testing tasks.

随着软件规模庞大和功能复杂化，软件测试的压力与日俱增，而回归测试并行化的理想方案受到物理资源限制。随着移动设备和网络技术发展，跨平台软件呈现普及趋势，然而多平台测试仍具有技术难度。云计算模式的测试服务是并行、多平台测试的解决途径，然而需要应对多场景组合爆炸、海量测试的执行效率、以及多软件错误的同步定位问题。 本项目从工业界问题出发，研究并行、多平台的云测试服务的关键技术--组合测试、测试用例排序和自动化错误定位。这些研究方向近年来很活跃，然而并非针对云测试而开展。从现代软件特点及软件测试问题出发，本项目在并行、多平台的云测试场景下对关键技术问题进行研究。 本项目研究提高回归测试速度、改善持续集成质量、增强错误定位效果的理论和方法;同时为云测试提供新的思路，有助于并行、多平台的测试实施。本项目将利用组合测试、回归测试和错误定位技术，针对现代软件特点及工业界问题实现云测试服务和相关工具。

现代软件具有“逻辑复杂、功能复杂、应用场景和平台多样”的特点，给软件测试带来巨大的压力。将回归测试并行化、软件测试多平台化是理想的解决方法，然而受到物理资源和实施难度的限制。本项目提出利用云计算模式的测试服务为并行测试和多平台测试提供解决方案。.为我们按照研究计划对申请书中所要求的四个方面进行了深入研究，还根据具体研究需要和进度，从理论分析、方法研究和应用研究三个角度对原有内容进行了相关扩展研究。按照预计的研究内容和其后补充的研究内容，分为以下几个方面:.☆研究目前现有的错误定位技术，分析不同因素对错误定位效果影响.☆研究程序信息与程序缺陷预测之间关系，并提出程序缺陷预测方法.☆研究测试用例优先性，缩小测试用例集.☆研究移动应用云测试，加速测试.同时在研究过程中，我们提出了一种创新性的错误定位方法：参考历史版本获取并削减程序结构对错误定位影响；两种移动测试方法：基于Gui可操作区间自动识别测试方法和基于Gui帧率检测程序流畅性的测试方法，.且已受理。.通过本项目研究，我们在以下四个方面取得显著进展：.☆基于程序谱错误定位以及缩小回归测试代价理论研究.☆基于程序信息缺陷预测以及程序缺陷检测方法研究.☆将上述研究结果应用到移动应用云平台测试.☆已提出方法有效性与可扩展性研究.相关工作在《IEEE Transactions on Software Engineering》、《Journal of Information and Software Technology》、《Journal of Systems and Software》、《IEEE Transactions on Reliability》、《Transactions on Services Computing》等杂志和《ACM International Symposium on Mobile Ad Hoc Networking and Computing》、《IEEE International Conference on Software Quality, Reliability & Security》等会议上发表论文。