Inferring dynamic credentials for role-based trust management

Abstract

The topic of this paper is the role-based trust-management language RT_0, a formalism inspired by logic programming that handles trust in large scale, decentralised systems. We provide a purely operational semantics for the language in which credentials can be established using a simple set of inference rules. We then extend RT_0 to include time validity and boolean guards that control the availability of credentials. In such an extended framework, credentials are conditional on the availability of supporting credentials in the execution context. In addition to a set-theoretic and a logic-programming semantics, we develop for the extended language a series of increasingly powerful inference systems for establishing these conditional credentials. By means of simple but realistic examples, we demonstrate the expressiveness and usability of our language, warranting its integration into existing trust-management tools.