Optimal release time determination via fuzzy goal programming approach for SDE-based software reliability growth model

Abstract

Faults in software can be of varied severity and so is the priority with which they are fixed. Hence, these two factors impact the reliability of the software to a large extent and is worthy of being taken into consideration while framing the software reliability growth model (SRGM). In this paper, a stochastic differential equation-based SRGM has been formulated where detection of faults has been associated with severity of fault occurrence and the process of correcting them have been knit with priority of defect fixing. Another matter of concern is the determination of the optimal time to release the software in the market with high reliability figures keeping the development cost margin minimum. Release time determination in the literature is done mainly using crisp or type-1 fuzzy system approach wherein the cost constraint in the optimization framework is considered to be of deterministic nature which is something highly unrealistic. In this study, the cost constraint is considered to comprise of a randomized cost budget with variable coefficients being interval type-2 fuzzy numbers which can represent the dynamicity associated with changes in specification during software development process in a much more realistic fashion. This encourages the release time determination problem to be framed as a multi-objective chance constrained optimization problem, which is thereby solved using fuzzy goal programming approach based on Taylor series method. The proposed SRGM and the release time approach have been validated on real time data, which show much promising results than many of those available in the literature.