A NHPP Software Reliability Growth Model Considering Learning Process and Number of Residual Faults

Abstract

A large number of software reliability growth models have been proposed to analyze the reliability of software application during the testing phase. With the increasing demand to deliver high-quality software, more accurate software reliability models are required to estimate the optimal software release time and the cost of testing efforts. In this paper we firstly demonstrate that the G-O model based on NHPP doesn’t need to consider imperfect debugging and new mistakes introduction during the debugging process. Then on the basis of G-O model a flexible and accurate SRGM is proposed, which considers the time-dependent fault detection rate. And in this model the value of fault detection rate is not only calculated with the number of faults remaining in the software system, but also calculated with human’s learning process during testing phase. It is more realistic. Moreover, the experiment results show that the proposed model fits the public failure data better and can provide more accurate software reliability prediction compared with other existing models.