Testability of the Safety Software based on Specification in Statechart

摘要

For hardware design, the testability is considered at the design stage to make the testing of the faulty components easy. The hidden faults in the software are exposed as certain conditions are met. The problem with software testing is that it is not easy to decide when enough testing have been performed. The stopping criteria for software testing have been either coverage criteria or certain reliability criteria. But if we can calculate the testability of the software, it can be used to support the stopping criteria, or to allow optimal resources to each software component according to their testability within limited resources. It can also be used to modify the software to have better testability. For safety system software, the software specification is increasingly prepared using formal methods having pre-defined notations with precise semantics. One of the methods that are widely used is the statechart that is based on the extended finite state machine (FSM). The statechart added concurrency, broadcasting, and hierarchy among the states to the extended FSM to enhance the descriptive power and avoid the state explosion problem. In this study, a method to evaluate the testability of software specification in statechart was proposed based on the number of test cases required, which could be used to evaluate the software in the early life cycle phase. The measure was applied for simple FSMs and simple hardware circuits.