Towards Software Product Lines Optimization Using Evolutionary Algorithms

Abstract Software product line (SPL) engineering methodology assist to create a range of software products within less time and cost but with high quality by the reuse of core software assets, which has been tested. Thus, testing is crucial for successfully deploying SPL. As the product features increases, testing process can be time-consuming. Testing in SPL is regarded as a combinatorial optimization problem. Evolutionary algorithms were reported to provide good results in such class of problems. This research provides a framework to compare different multi-objective Evolutionary Algorithms performance regarding software product line context. We report on the problem encoding, variation operators and different types of algorithms: Indicator Based Evolutionary Algorithm (IBEA), Strength Pareto Evolutionary algorithm II (SPEA-II), Multi-Objective Evolutionary Algorithms based on Decomposition (MOEA/D) and Non-Dominated Sorting Genetic Algorithm II (NSGA-II). The framework will provide preliminary results on different Feature Models (FMs) to measure their feasibility to optimize SPL testing.