Nonhomogeneous Poisson Process Software Reliability Growth Models And Applications

With the development of computer technology, software is used in almost every field of human's life to control and monitor the complicated system. So, more and more attention is paid on how to control software quality effectively. As an important part of software quality, software reliability measurably describes the faults that happen in software developing and using process. By evaluating reliability correctly, predicting how reliability improves with testing process, it can provide information about the time to release the software and the method to allocate testing resources. Then software can be released on time with fulfilling users'reliability requirement and reducing the cost of software.As an important sort of software reliability model, Non-Homogeneous Poisson Process Software Reliability Growth Model (NHPP SRGM) is a quite successful tool in software reliability engineering and the most suitable and easiest model for software reliability evaluation. It is a research hotspot in software reliability engineering that how to improve the evaluation and prediction ability of NHPP SRGM.The research of this thesis includes:1) An improved NHPP model with time-varying fault removal delay is proposed. It is assumed that perfect debugging is an acceptable assumption (based on thorough regression testing) but instantaneous fault removal is not. Then, how fault removal time changes with the time when the fault occurs is analyzed. For the fault that is detected later, it would be more difficult to make sure where the fault locates, and would consume more time to remove, so fault removal efficiency would decrease. Then, an improved NHPP model with time-varying fault removal delay is proposed, it is validated by using some testing data. The results show that the improved model fits and predicts the data better.2) An embedded software reliability testing and evaluation tool (ESRTE) is constructed by incorporating the theory researched above into it. ESRTE is an integrated testing and evaluation system including these four functions: software usage model edit, test case generation and management, test process execution and tracing, software reliability evaluation and prediction. ESRTE tries to provide an automated software reliability testing and evaluation environment.