Distributed prognosis of discrete event systems under bounded-delay communications
Shigemasa Takai, Ratnesh Kumar · 2009
The task of failure prognosis requires the prediction of impending failures. This is possible if each failure trace possesses a nonfailure prefix for which at least one local prognoser is unambiguous that a failure in future is inevitable. This paper formulates and studies the problem of distributed prognosis of discrete event systems, where the local prognosers exchange their observations of the events executed by the plant for the sake of arriving at a prognostic decision. The observations are exchanged over communication channels that introduce bounded delays. A property of joint-prognosability is introduced to capture the condition under which any failure can be predicted by some local prognoser prior to its occurrence. We provide an algorithm to check the joint-prognosability property that uses the plant model extended to include the communication channel models.