Distributed state reconstruction for discrete event systems

Abstract

We consider the state estimation problem for stochastic discrete event dynamic system (DEDS) obtained by the parallel composition of several subsystems. A distributed inference algorithm is developed in the case of distributed observations. It is composed of asynchronous agents that only have a local view of the model and of observations. This algorithm only handles local states of subsystems, which is a way of avoiding the state explosion difficulty of large concurrent systems.