An Approach to Clustering Feature Model Based on Adaptive Behavior for Dynamic Software Product Line

Dynamic Software Product Line (DSPL) is intent to support adaptive software system to meet requirement changes and evolving resource constraints during runtime. The adaptation may be accomplished by reconfiguring adaptive behavior at adaptive point in feature model that describes variability of system. The decision making of dynamic variability management for variation point of feature model is challenges in DSPL. This research proposes an approach to clustering feature model on adaptive point based on adaptive behavior represented with adaptive context. An approach for similarity uses Fuzzy clustering and Local Approximation of Membership (FLAME) algorithm to reconfigure software system. The MAPE-Kc framework is used for adaptive task operation in order to reducing adaptation time of decision making process in DSPL. The effectiveness of the approach is demonstrated with a case study.