Automated Planning for Self-Adaptive Systems

Self-adaptation has been proposed as a viable solution to alleviate the management burden that is induced by the dynamic nature and increasing complexity of computer systems. In this context, architectural-based self-adaptation has emerged as one of the most promising approaches to automatically manage such systems, resorting to a control loop that includes monitoring, analyzing, planning, and executing adequate actions. This work addresses the challenges of adaptation planning -the decision-making process for selecting an appropriate course of action- with a focus on the problem of provisioning automated mechanisms for assembling adaptation plans, as a means to enhance adaptive capabilities under uncertainty. To this purpose, adaptations are modeled in a hierarchical manner, defining primitive actions, guarded actions, and deliberate plans, which may guide the system towards a desired state.