An architecture-centric approach for multi-agent system development and application

The era of distributed software environments is emerging and research on multi-agent systems (MAS), which tries to solve complex problems using entities called agents, is on the rise. This paper proposes an architecture-centric method for developing MAS that supports the important phases of systematic software development. In particular, this approach is geared towards supporting systems' properties specially focused on agents' coordination and autonomy. In order to support agent coordination and autonomy, a goal-based approach is utilized for the problem domain analysis, and individual agents are then mapped to the system's refined goals. Further, architectural styles and patterns are applied to generate the overall design of MAS. UML (Unified Modeling Language) and ADL (Architecture Description Language) are used for modeling and formalizing the MAS architecture. The proposed architecture is applied to ITS (Intelligent Transport Systems) domain and a proof-of-concept prototype has been developed to validate our approach.