The \pi-calculus is a formalism of computing in which we can compositionally represent dynamics of major programming constructs by decomposing them into a single communication primitive, the name passing. This work reports our experience in using a linear/affine typed \pi-calculus for the analysis and development of type systems of programming languages, focussing on secure information flow analysis. After presenting a basic typed calculus for secrecy, we demonstrate its usage by a sound embedding of the dependency core calculus (DCC) and by the development of a novel type discipline for imperative programs which extends both a secure multi-threaded imperative language by Smith and Volpano and (a call-by-value version of) DCC. In each case, the embedding gives a simple proof of noninterference.

We introduce a typed π-calculus where strong normalisation is ensured by typability. Strong normalisation is a useful property in many computational contexts, including distributed systems. In spite of its simplicity, our type discipline captures a wide class of converging name-passing interactive behaviour. The proof of strong normalisability combines methods from typed l-calculi and linear logic with process-theoretic reasoning. It is adaptable to systems involving state and other extensions. Strong normalisation is shown to have significant consequences, including finite axiomatisation of weak bisimilarity, a fully abstract embedding of the simply-typed l-calculus with products and sums and basic liveness in interaction.

Abstract. Exploiting linear type structure, we introduce a new theory of weak bisimilarity for the π-calculus in which we abstract away not only τ-actions but also non-τ actions which do not affect well-typed observers. This gives a congruence far larger than the standard bisimilarity while retaining semantic soundness. The framework is smoothly extendible to other settings involving nondeterminism and state. As an application we develop a behavioural theory of secrecy in the π-calculus which ensures secure information flow for a strictly greater set of processes than the type-based approach in [20, 23], while still offering compositional verification techniques. 1

Nobuko Yoshida Imperial College London ABSTRACT We introduce a new expressive theory of types for the higher-order p-calculus and demonstrate its applicability via non-trivial security analyses of a simple class-based language with distributed code mobility. The new theory significantly improves our previous one presented in [52] by the use of channel dependent/existential types. New dependent types control dynamic change of process accessibility via channel passing, while existential types guarantee safe scope-extrusion in higher-order process passing. This solves an open issue in [52], leading to significant enlargement of original typability. Two basic security concerns for mobile computation, secrecy for data confidentiality and access controls for authorised resources are analysed in a uniform type-based static framework, culminating in the noninterference theorem and authority-error freedom in the presence of higher-order code mobility. The generality and expressiveness of the new type discipline are tested with a sound embedding of multi-threaded class-based language with dynamic code/class distribution, enforcing secrecy and accessibility.