Coordinating Activities and Change: An Event-Driven Architecture for Situated MAS

Agent activities and environment change are what make things happen in a multi-agent system (MAS). Complexity in a MAS comes from non-trivial dependencies between activities (social interaction), and between activities and environment change (situated interaction). As they are used to manage social (agent-agent) dependencies, coordination artefacts could also be used to govern situated (agent-environment) dependencies.

Along this line, in this paper we propose an event-driven architecture for complex MAS that exploits coordination to handle all sorts of dependencies in a uniform way. We first motivate the underlying meta-model and put some well-known agent-based frameworks in perspective, then we discuss its general articulation along with its reification within the TuCSoN coordination middleware.

keywords   MAS architecture; event-driven model; situatedness; coordination; TuCSoN

journal or series

 Engineering Applications of Artificial Intelligence (EAAI)