Dynamic high-level in self-adaptive systems

An adaptive software has the ability of modify its own behavior for responding at runtime to changes in its surround context. The ability to self-adapt to changing operational contexts or scopes means that the system can modify its behavior with respect to its original requirements at run time. Recently a new dimension of adaptation is increasing its relevance for complex systems: the ability to accommodate changes in the original requirements taking place when the system is operational. A self-adaptive system should also, to some extent, enable dynamic reconfigurations to meet new or modified requirements without the need of a new design/develop/test/deploy cycle. In this paper we present a software architecture exploiting an explicit, actionable requirement model to drive self-adaptation activities. The architecture includes some semantic-based models able to manage requirements at runtime. We show an implementation of this architecture that leverages the semantic representation of changes to requirements at runtime. We show how the requirements models are related to the software architecture and how runtime changes in these models can influence the adaptive behavior of the system.