Sensitivity Analysis for Cost Optimization of Software Reliability Growth Model with Multiple Change Points

Abstract

Purpose: It is recommended to complete the testing of a software and release it to maintain its market growth, and lower development research costs are critical for software developers. In this paper, we consider the optimal release time to be achieved when the program development cost is minimized in a model that has a bathtub-shaped failure detection rate with a change point. Here, we identify the various parameters that significantly affect the release time.BRMethods: Considering the cost function of the average failure function for this model, we aimed to estimate the release time to minimize it, to observe the changes in the release time when a parameter changes around the time of release, and to find a parameter that significantly affects the release time.BRResults: The values of the parameters α, λ, Φ and γ₁ continuously increased, and the total cost C(T) decreased and affected the release time T . Further, as the values of the parameters and increased, the total cost γ₃ and release time C(T) do not appear to have an effect.BRConclusion: In this paper, the optimal release time was obtained by considering the cost function for the reliability growth model with a bathtub-shaped failure detection melting with a change point, and the effect on the release time was investigated as each parameter changed. Since the parameters formulated based on the initial data collected are generally inaccurate, it is important to determine the sensitivity.