A model abstraction method for networked discrete-event systems

Abstract

Networked discrete-event systems consist of autonomous subsystems that utilise digital communication for solving cooperative control tasks. Each subsystem is composed of a local agent connected to a network unit, which comprise a synchronisation unit and a decision unit. This paper introduces a synchronisation unit that enables the local agents to reach target states autonomously (tracking control) and to execute synchronous state transitions with other subsystems. For designing the decision units, this paper proposes a compositional abstraction method for networked discrete-event systems to obtain a reduced model that describes the cooperative behaviour of the subsystems defined by the synchronous state transitions. If a sequence of synchronous state transitions appears on the subsystem's state trajectory into its target state, then the decision units have to adapt the state trajectory to execute a deadlock-free order of these synchronous state transitions. As the decision units have only local model information, they exchange their reduced abstracted models among each other to locally compose the abstracted overall model. It is proved that each local decision unit can determine the desired execution order of synchronous state transitions as a state sequence in this composed abstracted model.