A Formal Approach for Modeling Software Agents Coordination

In this paper, we propose a nested Petri net approach to model the coordination of multi-agent systems. A higher level net (called host net) defines the movements and coordination mechanism of agents, while lower level nets (called agent nets that are tokens of the host net) model the behavior of individual agents. A dynamic channel concept and the notation are introduced for modeling the communications and coordination between the host net and agent nets. We demonstrate our modeling approach through an e-market example. Our approach promotes the extensibility and flexibility of multiple agent system design through dynamic channels.