Application Of Change-point Method In The Research On Software Reliability Models

With the increasing of society informatization, social daily operation depends on software system more and more. However, the capacity of failure allowed in these systems begins smaller more and more. What's more, the improvement of hardware reliability level makes software quality, especially software reliability being a problem that can't be ignored. In most system, the worst execution effects are caused by software's unreliability. To improve the software quality, software reliability engineering plays an important role in many aspects throughout the software life cycle.On the aspect of software reliability model and its evaluation, there have been some software reliability models, and these models described software reliability in some degree. However, more or less limitations exist in all kinds of models, and there is no model that can be used widely in an exact way. Therefore, people keep on exploring models to evaluate software reliability. In order to achieve the requirement of improving software quality, a change-point model for software reliability is discussed.In this paper, we consider the software reliability growth model that incorporates with change - point problem. The proposed model utilizes the failure data collected from software development projects to analyze the software reliability and the remaining errors of a released software program. We investigate the new multi-model and demonstrate its applicability in the software reliability engineering.Our analysis suggests that if a change-point exists in a testing process, it should be considered in creating a software reliability estimation model. Based on the probing of theories, this paper applies the new multi-model to the statistics of the datasets often experience changes, which verifies that the mixture model is better than any of its component models and more useful in enhancing models' precision.