Matrix Approach for Verification of Opacity of Partially Observed Discrete Event Systems

Abstract

Opacity is a confidential property characterizing whether the secret of a system can be inferred or not by an outside observer (also called an intruder). This paper focuses on presenting a matrix-based approach for verification of opacity of nondeterministic discrete event systems (DESs). Firstly, the given system is modeled as a finite-state automaton. Further, based on Boolean semi-tensor product (BSTP) of matrices, the algebraic expression of the observable dynamic of the system can be obtained. We, respectively, investigate current-state opacity and K-step opacity owing to the equivalence between a few opacity properties. Finally, necessary and sufficient conditions are presented to verify whether the secret is opaque for a given system, and the proposed methodology is tested effectively by examples. The matrix-based characterization of opacity proposed in this paper may provide a helpful angel for understanding the structure of this property.