An Empirical Analysis of the Utility of NHPP SRGM

Wenqian Jiang,Ce Zhang,Wenyu Li,Miaomiao Fan,Zhichao Sun,Yafei Wen,Kaiwei Liu

DOI: https://doi.org/10.1109/icnisc54316.2021.00112

2021-01-01

Abstract:The software reliability growth model SRGM is an effective tool for evaluating software reliability in the software testing phase. The SRGM based on the non-homogeneous Poisson process NHPP is the most widely used model. So far, the validity range of the model has not been known. This article first reviews the construction process of the classic G-O model, summarizes the modeling process of NHPP SRGM, and briefly analyzes the research content and classification of SRGM. In order to verify the fitting and prediction performance of many NHPP-type SRGMs in the actual test environment, this paper uses 26 SRGMs to conduct experiments on 19 real data sets, and analyzes in detail the performance of 11 imperfect debugging models on 11 data sets. Fitting curve and fitting index value, as well as the prediction curve for predicting the number of failures of the software in the future. The results of the example verification show that no model can have excellent fitting and prediction performance in any test environment, that is, the performance of the model in the application environment is different and not universal.