Optimal software testing-resource allocation with operational profile: computational aspects

Abstract

Software testing-resource allocation plays a significant role in software project management. During the last two decades, a number of software-resource allocation problems have been formulated from various points of view, by taking account of several optimization criteria with constraints. First, we consider three optimization problems with constraints in a combination of expected total software cost, cumulative software testing effort and desired software reliability level. Especially, we focus on the operational profile of a software product and formulate rather different software testing-resource allocation problems from the existing ones. The optimization algorithms to determine the software testing resource for each software component are given by means of nonlinear programming. Next, we consider a statistical estimation problem of the optimal software testing-resource allocation, where a simulation-based parametric bootstrap method is applied to obtain not only the confidence interval, but also the higher moments of an estimator of the optimal software testing effort for each software component. In simulation experiments, we give an illustrative example on how to evaluate the uncertainty of the estimator of the optimal software testing-effort allocation.