The Research Of Software Reliability Model With Test Effort

With the rapid development of computer science, computer software plays an increasingly important role in our modern life. Software quality is a crucial guarantee to its existence, and software reliability, an important criterion to its quality. Now there aren't reliability models which are widely applied among many software reliability models. For the same software, different hypothesizes may be correspond to different reliability models. Whether models are successful those are closely contacted with true situation. So research of software reliability models focus on how to make reliability models accord with actual situation.Scientists have been proposing types of models to predict the software failure since the first software reliability model was proposed. Currently there are more than one hundred software reliability models, among which NHPP models take up a rather large percentage.This dissertation firstly introduces the research meaning, origin and development at home and abroad, presents some basic concepts and theories about software reliability and SRMs along with evaluation standards of SRMs and concretely analyzes several classic SRMs. Then, some released NHPP software reliability models are carefully studied, and we use some more rational assumptions and make some improvement to NHPP software reliability models. In the test process, the consumption of resources (human, material and financial resources) in test process, test environment, and the input level of tester can affect the results of test. The consumption of resources in test process will vary gradually through time. We describe the consumption of resources with test effort. Based on this, perfect debugging and imperfect debugging in the test process are also considered, and a new NHPP model with respect to test effort is proposed. After contrasting with some released models, the new model shows better performance to fit data and theoretically can be applied into projects.