An Approach for Optimized Feature Selection in Software Product Lines using Union-Find and Genetic Algorithms

Abstract

In Software Product Line (SPL), feature model is highly recommended to manage the commonalities and variability of features under resource constraints of mandatory, optional and alternative. Features with mandatory constraints and high in dependency with other features are identified as crosscutting concerns; reduce the reusability of resources. It is important to find and modularize these concerns at modeling level. With this practice, these concerns do not effect if deletion or addition is required from entire system. In this paper we have applied Union-find algorithm to find crosscutting concerns in feature model. We evaluated our approach by applying on an automobile feature model with various dependencies between features, and found required crosscutting concerns. By this approach, identification of crosscutting concerns and their modularization made easier. Further, we have also applied genetic algorithm to get optimized feature selection under cost constraint with high performance. In SPL, as crosscutting concerns are mandatory features with fix cost and performance, optimization on feature model is necessary under consideration of crosscutting concerns. Our approach found all possible products according to crosscutting concerns, cost and performance at modeling level of an automobile feature model. At last, we found all products from minimum to maximum cost with respect to least maximum performance by using GA optimization technique.