Soft-Goal Approximation Context Awareness of Goal-Driven Self-Adaptive Systems

In goal-driven self-adaptive systems, a goal model is used as a requirement model and is held by the system even at runtime. At this moment, the self-adaptive system, which can change its behaviour at runtime, will be able to reason over the variability within the goal model. It will then be able to find the best behaviour to deal with environment evolutions. However, the uncertain nature of the requirements engineering concepts in a real dynamic world is not always rightfully defined. In particular, quality requirements runtime changes are almost never considered in the literacy. Specifically, a problem we highlight here is the lack of context consideration in quality constraint approximation. Our purpose is to clearly define this problem and to propose a first solution. In this paper, we introduce a flexible version of the quality constraints. This new kind of quality constraints can be rewritten at runtime to tackle the context change induced by the environment change. To stick the constraint definition to the context change, we developed a new algorithm which modifies the specification of the quality constraints at runtime.