CSR---EHS: Collaborative: Directed Real-Time Testing

该项目正在为实时嵌入式系统开发覆盖率驱动的定向自动测试生成的新方法。这项研究既涉及自动测试生成的理论基础，也涉及实时嵌入式系统的实用测试和覆盖工具。许多影响系统行为的信号可能无法观察到，或者可能很难在不影响系统行为的情况下在线测量，并且系统本身可能不容易重置到初始状态以驱动不同的测试。在本研究中，系统中的任务被建模为事件自动机，联合事件自动机表示系统中所有任务的事件自动机的组成。这项研究是探索游戏和控制理论的技术，从一个自动构建的抽象交互图的实现中合成一个测试导演器。测试导向器在系统的执行中引入可变延迟，从而满足最大数量的覆盖目标。测试工具DIRECT实现了嵌入式系统的直接实时测试，将静态分析(交互图构造)、控制理论(测试指挥构造)和动态分析(在测试指挥在场的情况下进行在线系统监控)相结合，实现了高交互覆盖率。特别是，这些问题将在资源受限的环境中研究，其中在线测量和测试导向器实施都必须在时间和空间上进行优化。这些工具和技术将在我们可用的两个试验台上进行评估：一个是由加州大学圣克鲁斯分校的一名PI开发的乐高头脑风暴机器人之上的嵌入式开发平台，另一个是加州大学圣克鲁斯分校开发的用于海洋跟踪的感应板。

This project is developing new methods for coverage-driven directed automatic test generation for real-time embedded systems. The research addresses both, the theoretical foundations of automatic test generation, and practical test and coverage tools for real-time embedded systems. Many signals affecting system behavior may not be observable, or may be difficult to measure online without affecting system behavior, and the system itself may not be easily reset to an initial state to drive different tests. In this research, tasks in the system are modeled as event automata, and the joint event automaton represents the composition of event automata for all tasks in the system. The research is exploring game and control-theoretic techniques to synthesize, from an automatically constructed abstract interaction graph of the implementation, a test director. The test director introduces variable delays in the execution of the system such that a maximal number of coverage goals are met. A testing tool, DIRECT, implements directed real-time testing for embedded systems, combining static analysis (interaction graph construction), control theory (test director construction) and dynamic analysis (online system monitoring in the presence of the test director) to achieve high interaction coverage. In particular, these problems would be studied in a resource-constrained setting, where both online measurements and test director implementation must be optimized for time and space. The tools and techniques would be evaluated on two testbeds available to us: an embedded development platform on top of Lego Mindstorm robots developed by one of the PIs at UC Santa Cruz and a sensor board for marine tracking developed at UC Santa Cruz.