Binary multi-objective PSO and GA for adding new features into an existing product line
Abdullah Al Mamun, Fahim Djatmiko, Mridul Kanti Das · 2016
Software product line evolution is a decision-making problem where it is determined that which features are the best candidates for the different product. In this paper, we analysis two popular optimization technique for finding qualified features for next release. In this case, the qualification of features is determined by the product value, product integrity, and AND dependencies between features. This study may help a software manager in feature selection. For this purpose, first, we encoded the feature model into a binary string. Then we able to find possible alternatives by using binary multi-objective genetic search algorithm NSGA-II and particle swarm optimization PSO. From these options, the product line manager can quickly select the best-fitted solutions according to the requirements. In this implementation, a feature model called ONLINE ELECTRONIC SHOPPING from Software Product Line Online Tool SPLOT is used where the total number of features is 290 assuming that 136 features are already implemented in the existing software. Finally, 30 out of 154 features are selected successfully in five product variants. The empirical result shows that GA performs better than PSO.