Research Of Automated Testing Techniques For Safety-critical Software

Safety-critical software usually refers to the software that used in the medical area, railway transportation, aerospace and other fields. This kind of software always has a complex structure and a high level of quality requirement. The malfunction of the software may lead to disastrous consequences. So it is necessary to go through enough tests of the software to avoid accidents. However, manual testing will waste too much time and money. Thus, the research of automated testing of safety-critical software would be of great significance.Unlike conventional software, safety-critical software is not only a huge system, but also has special requirements for security and reliability. Therefore, appropriate methods of automated testing will improve the efficiency of the software testing as well as the security of the safety-critical software. However, due to the complexity of safety-critical software, most automated testing methods are not applicable to this kind of software. And some research achievements in this area are too complex for practical application. Therefore, this article researches the technology on automated test case generation and executing of safety-critical software.First of all. currently popular methods of software testing are discussed in two aspects: automated test case generation and automated test case executing. The advantages and disadvantages of these methods are analyzed. According to the SpecTRM toolset, user-defined syntax rules for software specification is especially introduced and developed in this article. Date-driven automation test framework is adopted to accomplish the whole automated testing process.Then, a method of rewriting software specification is proposed. This method, which is based on the state-machine model, has simple syntax rules familiar with Chinese grammar. Thus enables people to understand and use this method easily without having to learn too much extra mathematics knowledge. Focused on the special requirements of safety-critical software on reliability and safety, a theory of state mutation is discussed in the paper to generate test cases with fault logic as many as possible. Because of the complexity of the safety-critical software, data and logic of the test case are separated. After the test logic is automatically generated, data configuration will be executed to generate the test case sets. At last, taking the ATS subsystem in CBTC as an example, this article develops a set of tools to automatically test the programming code and user interface. The coverage rate can reach as much as90%. The tool set can also evaluate the safety of the ATS subsystem properly. In addition, a three-dimensional visual and interactive system is developed to satisfy the requirements of real-time display and testing of the system, which will obviously reduce the dependence on real environment when testing the software and save money.