Research On Software Reliability Evaluation Technology Based On Markov Chain

With the development of Object-oriented software technology, reliability evaluation of object-oriented software, as an important part of highly reliable software, is concerned strongly by scholars. It has important significance for optimizing software architecture, improving software development process, guiding software testing. In this paper, the software reliability evaluation technology and methods based on Object-oriented software Markov chain usage model are studied, which can be used to evaluate the software reliability in software analysis and the testing phase. Briefly, the main work of this paper includes:(1) In order to solve the problem that the existing software reliability evaluation method is fragmentation, a software reliability evaluation framework based on Markov chain usage model has been constructed. The role of the framework has been pointed out, and according to the framework, the flow of reliability evaluation has been presented.(2) In order to solve the problems that state space of the existing usage model is too large and model description is too difficult, a method has been proposed, through which the scenario-level and use case-level software Markov chain usage model can be established from UML model. The generation algorithm of usage model and the verification algorithm for extended UML model are presented. As an example, this paper details the methods and procedures of ATM usage model generated from the extended UML model.(3) During the generation process of usage model, no method is given to determine the transition probability. Hence, a method is proposed, through which the transition probability of Markov chain usage model is determined by AHP. After the study of AHP, we proposed a method to determine the transition probability of Markov chain usage model, which is of great benefit to calculate the software reliability value in software early reliability evaluation stage. Finally, the method of ATM usage model transition probability is expected by the proposed method.(4) A software early reliability evaluation method based proposed Markov chain usage model has been presented. According to Markov chain usage model, we establish a scenario-level early reliability evaluation model. The reliability expecting algorithm of Object-oriented software is obtained from the definition of reliability, which helps to finish software early reliability evaluation. The early reliability of ATM is expected by the presented method too.(5) A software reliability growth model (SRGM) based discrete NHPP has been presented by researching and extending basic model. We consider imperfect debugging and new fault derived and changed with fault debugging of our assumption, which is more realistic, extend basic model and establish a SRGM based discrete NHPP. The predictive power and goodness-of-fit of the proposed models are examined by three public failure data sets. The results demonstrates that compared with other existing models, the proposed model has better prediction and goodness-of-fit.In this paper, we study the methods and technologies of Object-oriented software reliability evaluation. The methods and technologies can provide theory and decision support for software reliability evaluation system design and implementation, and it also can help to assess the software reliability comprehensively and improve software quality.