Model for situated multi-agent systems with regional synchronization

In this paper we present a generic model for situated multi-agent systems (MAS). Situated agents live and act in the present: they perform situated actions, i.e. actions selected on the basis of their position, the state of the world they perceive and limited internal state. The model we discuss is based on the model of influences and reactions to influences developed by J. Ferber and J.P. Muller. This model separates what agents do and the reaction of the environment upon this, solving the fundamental problem of handling simultaneous actions. However, the model is limited to synchronous descriptions of MAS evolution. Forcing all agents of the MAS to act in lock step does not fit with autonomy of agents and scalability. Besides, global synchronization implies centralized control, in general an undesirable property of MAS. The model we present is asynchronous. It does not impose all agents to act at one global pace, but instead agents themselves decide when to perform their next actions. The model supports simultaneous actions through "regional synchronization". With regional synchronization agents form synchronized groups on the basis of their actual locality. Different groups act asynchronously, while agents act synchronously in their group. The result is a model that does not suffer from the drawbacks of global synchronization while it preserves the properties for handling simultaneous actions.