Relative Codiagnosability of Decentralized Failure Diagnosis Systems

Abstract

The notion of relative diagnosability has been proposed to capture the feature of diagnosable discrete event systems in a centralized framework. As a continuation, this paper addresses the relative codiagnosability of failure event occurrences under the decentralized framework. Roughly speaking, a discrete event system equipped with  local independent diagnosis components which are employed to deal with the local observation is called relatively codiagnosable provided that there exist some failure events which can be diagnosed based on at least one local observation of the system. Intuitively, relative codiagnosability is weaker than codiagnosability introduced by Qiu, W.. A necessary and sufficient condition for relative codiagnosability is presented and a polynomial-time algorithm to test the relative codiagnosability is developed by introducing opacity. Furthermore, some examples are provided to illustrate the presented results.