Prognosis of Patterns in Networked Discrete-Event Systems with Communication Delays

Abstract

This paper investigates the problem of prognosis of patterns in the context of networked discrete-event systems with communication delays. In this problem, the pattern to be predicted describes event sequences modeled by a finite state automaton; communication delays are delays in the observation channel and the delays are random and bounded. Given a pattern, in order to check whether an observation-delay system is predictable with respect to the pattern and predict the pattern k steps before it occurs, a new notion of k-step pattern predictability is defined and an off-line algorithm is presented for verifying the property. For a predictable system, a predictor is constructed to predict the occurrence of a pattern on-line during the evolution of the system.