Context-aware agents for developing AmI applications

In a layered vision of Ambient Intelligence, the intelligent services layer presents to researchers a great number of challenges, and it is especially interesting from the perspective of Artificial Intelligence and Multi-Agent Systems. While software agents have been used before for building AmI middleware, this paper proposes an agent-based architecture that is both scalability-oriented and context-aware. This research is based on two previous approaches by the authors: one dealing with context-aware information sharing in a system using self-organization mechanisms, and the other defining an implicit representation of context by means of agent hierarchies. In this paper, elements of both approaches are used to build an architecture in which agents are structured in a topology that relates to several types of context, and that can be used for context-aware information sharing and searching.