Diagnosability verification using compositional branching bisimulation

Abstract

This paper presents an efficient diagnosability verification technique, based on a general abstraction approach. More specifically, branching bisimulation including state labels with explicit divergence (BBSD) is defined. This bisimulation preserves the temporal logic property that verifies diagnosability. Based on a proposed BBSD algorithm, compositional abstraction for modular diagnosability verification is shown to offer a significant state space reduction in comparison to state-of-the-art techniques. This is illustrated by verifying non-diagnosability analytically for a set of synchronized components, where the abstracted solution is independent of the number of components and the number of observable events.