Research On Process Model Of Testing And Evaluation For Safety-critical Software

Software was applied in many areas, and always serves as central controller of intelligent system. Failures, especially those happened in software used in critical area, caused a great loss to users and society. Therefore, strict testing and evaluating (T&E) should be executed to this so-called safety-critical software (SCS). On the other hand, there're some deficiencies in the current T&E methods, such as delayed starting time, not discriminating the differences between functions and operations, which flubbed the dependability of the result.After the fundamental theories and existing researches relative to software reliability were introduced, a new framework of T&E method adapts for SCS was addressed then. It ensured software safety technology be emphasized along the whole course of T&E process. Besides that, it also attempted to replace testing time by test coverage as the benchmark of software reliability improvement for the sake of enhancing the accuracy of method and its result.The main contributions of this dissertation are as following:(1) According to the trait of SCS, dissertation raised a new software reliability-growth model (SRGM) based on component-partition and test coverage. This model renovate current SRGMs'status that counting reliability improvement through testing time, and turns to calculating whether test case could improve coverage of the faults. Therefore, result of the evaluation became more accurate when we using this model.(2) The lecture presents a whole new strategy about how software safety technology was involved into software reliability testing. A risk-related operational profile (RROP), including criticality and accelerative testing, was constructed. It suggests that those operations and functions with higher criticality deserve more testing investment when we take testing cost into account. Modeling method using UML that includes testing constraint and time marking was combined to the construct process. This technique decreases the probability of misunderstanding between test engineer and developer, and improves the availability of test cases which were generated from the RROP.(3) This dissertation proposed a new T&E process model adapt to SCS, which provides support to the whole testing and evaluating lifecycle. It overcomes those weaknesses of traditional process, and improves the testing efficiency and veracity of the result.The author carried out a comprehensive study about T&E framework, and delivered a practical method. Some experimental results illustrated that the new method could enhance the accuracy and efficiency of the testing and evaluating process.