Context-Awareness in Multi-agent Systems for Ambient Intelligence

Andrei Olaru

Abstract: There is a large body of research that lies at the intersection of the domains of context-awareness, multi-agent systems (MAS) and Ambient Intelligence (AmI)/Ubiquitous Computing (UbiComp). This is because, while multi-agent systems are an appropriate architecture for AmI implementations, one essential requirement for AmI is to be aware of the user’s context and to act accordingly. In order to implement context-awareness in a MAS for AmI applications, one must on the one hand choose an appropriate representation for context, that is suitable for agents of all sizes and functions, and, on the other hand, create an agent-based architecture that facilitates communication between agents that share context. This chapter presents a model, mechanisms and methods for integrating context-awareness in multi-agent systems for AmI. The model is based on experience with several implementations of MAS dealing with various aspects of context-awareness.