Requirements trade-offs analysis in the absence of quantitative measures

Abstract

Simultaneously satisfying multiple interacting and possibly conflicting software requirements is challenging. Quantitative cost-benefit analysis of alternative solutions is often hard or biased, and early decisions based on numerical estimates of requirements satisfaction are thus unreliable. We propose a trade-off analysis method that assists decision making in the absence of numerical data. We structure the requirements trade-off problem in terms of a goal model containing alternative design solutions and decision criteria. We propose a trade-off analysis algorithm that takes pair-wise comparisons of alternatives and determines the best solution among alternatives. The optimum alternative is decided by using a heuristic method, which may need to consult with domain experts. We take advantage of the Even Swaps method [1] to incorporate stakeholders' preferences into the decision analysis. The algorithm is implemented in a prototype tool and evaluated in an industrial case study.