Measuring Opacity for Non-Probabilistic DES: a SOG-Based Approach

Abstract

A system is opaque w.r.t. a secret and an observation map if for every run that leads to the secret, there exists at least one equivalent run which does not. The dichotomy of this definition, however, fails to measure just how much information the system keeps hidden. In fact, a system can be deemed opaque due to a single non-secret run, among an overwhelming number of secret equivalents. In this paper, we wish to redeem this drawback by the means of quantifying opacity into an opacity degree measuring the security of a system. After we formalize this numeral value for both the system and its corresponding Symbolic Observation Graph (SOG), we enhance our findings using a case study and some experimentation.