Decentralized Fault Diagnosis for Constant-Time Automata

Abstract

This letter investigates the verification problem of codiagnosability for discrete-event systems modeled by constant-time automata (CTAs). Codiagnosability requires that the occurrence of a fault event necessarily be detected by at least one local diagnostic agent within a finite number of events after the fault event occurs. In the present work, we first construct a verifier to check the codiagnosability of CTAs in the decentralized architecture. Additionally, an inappropriate, lower-complexity statement concerning diagnosability verification in our previous work is corrected. This correction prompts us to investigate the underlying reasons for increased complexity and to propose a condition for complexity reduction based on the verifier. Furthermore, another important contribution is our demonstration of the equivalence between diagnosability and codiagnosability for CTAs, an equivalence that does not hold in the framework of logical finite state automata. This equivalence serves as another verification technique for the codiagnosability of CTAs.