F4Plan: An Approach to Build Efficient Adaptation Plans

Today’s society increasingly depends on software systems subject to varying environmental conditions imposing that they continuously adapt. A dynamic adaptation reconfigures a running system from a consistent state into another consistent state. To achieve this goal, a reconfiguration consists in executing a set of actions leading from source to target configuration. The planning of actions has often been neglected in adaptation mechanisms, leading to naive sequential schedules statically predefined. EnTiMid, a ubiquitous software system for assisted living, is one of these adapting systems using basic adaptation plan. This situation may cause problems when considering adaptations involving large set of actions and/or devices, particularly for distributed service-based applications. We propose a framework to ease the integration of different planning algorithms that produce more efficient adaptation plan than an ad-hoc algorithm.