A method to optimize the scope of a software product platform based on end-user features

Abstract

Due to increased competition and the advent of mass customization, many software firms are utilizing product families - groups of related products derived from a product platform - to provide product variety in a cost-effective manner. The key to designing a successful software product family is the product platform, so it is important to determine the most appropriate product platform scope related to business objectives, for product line development. Selecting the optimized product platform scope is a challenging issue that must be handled carefully when designing a family of products. This research proposes a novel method to find the optimized scope of a software product platform based on end-user features. The proposed method, PPSMS (Product Platform Scoping Method for Software Product Lines), mathematically formulates the product platform scope selection as an optimization problem. The problem formulation targets identification of an optimized product platform scope that will maximize life cycle cost savings and the amount of commonality, while meeting the goals and needs of the envisioned customers’ segments. To measure the commonality within a product family, a software commonality index is proposed as an analytical tool to measure the amount of feature sharing among a family of software products. Analytical tools developed in manufacturing domain (i.e. Commonality Indices (CI)) have been analyzed and validated, through a linear regression analysis study, in order to propose this index. A simulated annealing based algorithm that can solve problems heuristically is then used to help the decision maker in selecting a scope for the product platform, by performing tradeoff analysis of the commonality and cost savings objectives. In a case study, PPSMS helped in identifying 5 non-dominated solutions considered to be of highest preference for decision making, taking into account both cost savings and commonality objectives. Our empirical results show that the...