A new evolutionary approach to decision-making in autonomic systems

Increasingly, autonomic systems are present in our lives. For this kind of systems the ability to self-reconfigure and adapt in response to changes in users requirements and environmental conditions is primordial. Several approaches have been proposed in the literature to achieve self-reconfiguration, however, as the complexity of the adaptive system grows, designing and managing the set of reconfiguration rules becomes difficult and error-prone. To tackle this limitation, we propose a new approach that uses a search-based evolutionary algorithm that explores valid configurations to find the most relevant one given a specific running context. Another salient advantage of our approach is the re-exploitation, in the context of adaptability, of the design knowledge and existing model-based technologies through the reuse of the feature model of the system.