Coordination Models for Multi-Agent Systems

The attempts to determine which features a software entity should present to be defined an “agent” at least agree on the fact that an “agent” is an autonomous entity which proactively interacts with its surrounding environment. A multi-agent system is then built by composing several independent computational activities, which have to interact with each other and with their environment. Therefore, a comprehensive view of a multi-agent system should not rely only upon the analysis of the internal behaviour of each agent, due to their intrinsically interactive nature. In fact, these systems are likely to exhibit a complex global behaviour, emerging from the mutual interaction among components, that is hard to be described and managed when communication is considered from a single agent's viewpoint.

In this paper, we claim that the design and management of a multi-agent system may take advantage from the choice and exploitation of a coordination model, that is, a high-level interaction abstraction aimed at globally ruling the behavior of the different system components. In order to illustrate the issues arising from the choice of a suitable coordination model for the design of multi-agent systems, we take Internet applications as a case study.