Expressing Security Properties Using Selective Interleaving Functions

Abstract

McLean's notion of Selective Interleaving Functions (SIFs) is perhaps the best-known attempt to construct a framework for expressing various security properties. We examine the expressive power of SIFs carefully. We show that SIFs cannot capture nondeducibility on strategies (NOS). We also prove that the set of security properties expressed with SIFs is not closed under conjunction, from which it follows that separability is strictly stronger than double generalized noninterference. However, we show that if we generalize the notion of SIF in a natural way, then NOS is expressible, and the set of security properties expressible by generalized SIFs is closed under conjunction.