Implementation-Oriented Mutation Testing of Statechart Models

Executable statecharts-based models of embedded real-time systems provide a solid basis for early detection of behavioral errors. However, there are two major semantic gaps between execution of system model and execution of its actual implementation. One is treating of time: simulated time (where every reaction to environment stimuli takes zero time) vs. running in real time. Another one relates to "maximum parallelism" of statechart components vs. actual concurrent tasks running under certain scheduler. As a result, test set that is adequate for system model might be non-adequate for system implementation. This paper presents a series of statechart mutations that allows to address the above differences and to evaluate "implementation oriented" adequacy of tests.