파레토 분포의 특징을 적용한 유한고장 NHPP 소프트웨어 신뢰성 모형에 대한 효용성에 관한연구

Abstract

SReflecting software reliability features during the software development process is an efficient development environment. The the finite failure NHPP for analyzing software failure patterns shows that the hazard function, which means failure rate, is constant, monotonically increasing or monotonically decreasing. In order to explore the efficiency of software reliability performance, this study proposed a model comparison of the efficiency considering the characteristics of the reliability model following the shape parameter of the Pareto distribution. In order to compare the efficiency of the model, meansquared error and decision coefficient were applied. To do this, a parameter estimation algorithm was performed using the maximum likelihood method and the numerical method. In the study, the failure analysis was performed using real data to evaluate the reliability model of the characterize the Pareto distribution.. The failure data trend analysis was performed by applying the intensity function and the mean value function related to the shape parameter, which is a characteristic of the Pareto distribution. To ensure data reliability, the Laplacian trend test was performed. The model that follows the shape parameter, which is a feature of the Pareto distribution proposed in this study, can be applied to various failure phenomena (more than 60% of the decision coefficient) and can be judged as an alternative model applicable in the software reliability field. Through this study, it can be concluded that software users can apply the shape parameter of various phenomena as a standard guide for evaluation and analysis of software failure trend.