New approaches to fault prediction and opacity enforcement of discrete-event systems

Abstract

In this work, we are concerned with developing new approaches to problems related to two discrete-event system properties, namely, opacity and predictability. Opacity is a property that ensures that a secret behavior of the system is kept hidden from an intruder, and, in this regard, the rst problem we address is how to make opaque a system which inherently is not (opacity enforcement). To this end, we propose a strategy which is capable of manipulating the state estimation of the intruder by shu ing event observations and also deleting some of them when strictly necessary. We also investigate how it is possible to mitigate the negative e ect of opacity enforcement on the capability of the legitimate receiver to accurately estimate the current state of the system. The second opacity problem addressed in this work is related to the main criticism of the opacity enforcement strategies existing in the literature, namely that in order to obfuscate the secret behavior of the system from intruders, some transmitted information is also concealed from the legitimate receiver. In this regard, we introduce the notion of (state-based) utility to refer to some crucial behaviors that must always be available to the legitimate receiver, even when opacity is being enforced. The last problem we address is fault predictability, where a system is predictable with respect to the fault event if, for every faulty behavior, we can be sure of the fault prior to its occurrence. We then present two strategies for the veri cation of disjunctive fault predictability (fault copredictability), one that is based on a diagnoser-like test automaton, and the second one that deploys veri ers. We also consider online fault prediction, present a strategy for designing local fault predictor systems, and address K-copredictability, namely, if all fault occurrences can be predicted at least K events prior to their occurrences