A process-algebraic approach for the analysis of probabilistic noninterference

Abstract

We define several security properties for the analysis of probabilistic noninterference as a conservative extension of a classical, nondeterministic, process-algebraic approach to information flow theory. We show that probabilistic covert channels (that are not observable in the nondeterministic setting) may be revealed through our approach and that probabilistic information can be exploited to give an estimate of the amount of confidential information flowing to unauthorized users. Finally, we present a case study showing that the expressiveness of the calculus we adopt makes it possible to model and analyze real concurrent systems.