基于执行序列的嵌入式软件时序异常检测

发布时间：2018-05-25 05:34

本文选题：接口自动机 + 时间约束　；参考：《计算机学报》2017年12期

【摘要】：时序特性是嵌入式软件的重要特性,实时嵌入式软件运行的正确性不仅依赖于任务执行结果,更依赖于任务执行时间.执行任务往往具有多个时间约束,且相互之间可能存在制约关系,对时间约束进行验证和确认是嵌入式软件测试的一个重要问题.文中提出一种基于执行片段的嵌入式软件时序特性检测方法(Trace-Based Temporal Defect Detection,TBTDD).一方面,基于扩展语义接口自动机(Extended Semantic Interface Automata,ESIA)模型,刻画嵌入式软件时序特性需求,分析不同时间约束间存在的相关关系类型,并提出基于相关矩阵的相关时间约束识别算法;另一方面,在目标软件运行环境中提取包含时间信息的执行片段,通过执行片段与时间约束模型的匹配,依据预先制定的时序特性检测准则,检验执行序列是否满足模型中各项独立和相关时间约束的要求,进而发现被测软件中存在的时序缺陷.实验以卫星定位系统软件为例进行建模与缺陷检测,并在执行片段集和缺陷检测能力等方面进行了对比分析.实验表明,该方法可有效检测软件运行过程中存在的各类异常时序,提高了软件时序测试的有效性和充分性.
[Abstract]:Timing is an important feature of embedded software. The correctness of real-time embedded software not only depends on the result of task execution, but also on the time of task execution. Execution tasks often have multiple time constraints, and there may be constraints on each other. Verification and validation of time constraints is an important problem in embedded software testing. In this paper, an embedded software timing characteristic detection method based on execution fragment is proposed, which is based on Trace-Based Temporal Defect Detection (TBT). On the one hand, based on extended Semantic Interface Automata (EA) model of extended semantic interface, the requirements of embedded software timing characteristics are described, and the correlation relation types between different time constraints are analyzed, and a correlation time constraint recognition algorithm based on correlation matrix is proposed. On the other hand, the execution fragment containing time information is extracted from the target software running environment. By matching the execution fragment with the time constraint model, the detection criteria are established according to the predetermined timing characteristics. To verify whether the execution sequence meets the requirements of each independent and relevant time constraint in the model, and then discover the time sequence defects in the software under test. The experiment takes the software of satellite positioning system as an example for modeling and defect detection, and makes a comparative analysis on the execution fragment set and defect detection ability. Experiments show that this method can effectively detect all kinds of abnormal timing in software running process, and improve the validity and adequacy of software timing testing.
【作者单位】：清华大学计算机科学与技术系;清华大学信息科学与技术国家实验室;波特兰州立大学电气与计算机工程系;
【基金】：国家自然科学基金(91218302,61472197) 北京市自然科学基金(4132062)资助~~
【分类号】：TP311.53

【相似文献】