Formal Aspects in Security and Trust

Abstract

We consider the problem of defining an appropriate notion of non-interference (NI) for deterministic interactive programs. Previous work on the security of interactive programs by O'Neill, Clarkson and Chong (CSFW 2006) builds on earlier ideas due to Wittbold and Johnson (Symposium on Security and Privacy 1990), and argues for a notion of NI defined in terms of strategies modelling the behaviour of users. We show that, for deterministic interactive programs, it is not necessary to consider strategies and that a simple stream model of the users' behaviour is sufficient. The key technical result is that, for deterministic programs, stream-based NI implies the apparently more general strategy-based NI (in fact we consider a wider class of strategies than those of O'Neill et al). We give our results in terms of a simple notion of Input-Output Labelled Transition System, thus allowing application of the results to a large class of deterministic interactive programming languages.