Multi-release software: A decision making mathematical approach for analysing the impact of infected patch

Abstract

Patching provides software organizations an option to handle the leftover faults and hence helps in maintaining the software. Here we are focusing on the problem where an organization releases the software early in the market and further releases software patches to improve the functionality of the software. Sometimes the patch released in the market might be the infected one which may increase faults and disturb the functionality of the software. The current work
presents the Generalised Modified Weibull (GMW) testing effort-based software reliability growth model that quantifies the effect of an infected patch on multi-release software. In our mathematical model, the leftover faults from the previous release and faults due to infected patches are taken care of in the next release. To develop a mathematical model, we have considered a scenario where a company releases a single patch between two consecutive releases. The model here is capable of estimating the hike in fault content due to the release of an infected patch. Genetic Algorithm is applied to estimate the optimal patch release time which minimizes the testing cost. Further, the model is validated on the real dataset to depict the validity of the proposed model.