Extending Strand Spaces for Analysis of Group Key Agreement Protocols

Abstract

Group Key Agreement enable participants to compute a shared secret key, from the secret con- newlinetributions of the participants. Modelling of Group Key Agreement protocols using conventional newline newlineformal models is challenging due to two reasons: The security requirements of Group Key newlineAgreement protocols are different from that of two-party protocols. Also, the assumption of newline newlineperfect cryptography, made for two-party protocols is invalid here because Group Key Agree- newlinement protocols rely on algebraic properties, which could be exploited by the attacker. Two of newline newlinethe security properties of Group Key Agreement protocols considered in this thesis are Partial newlineForward Secrecy and Implicit Key Authentication. Strand Spaces is a formal framework for newlinemodelling and analysing protocols. The thesis presents extensions to Strand Spaces to analyse newlinethe above security properties. newlineThe thesis proposes the notion of relative strength of Group Key Agreement protocols, newlinewith respect to Partial Forward Secrecy and a metric to compare the protocols with respect to newlinePartial Forward Secrecy. The proposal is illustrated on collections of Group Key Agreement newlineprotocols. newlineThe idea has been further extended to propose a computational trust model for dynamic newlinepeer groups. The model quantifies trustworthiness of individuals as well as groups on the basis newlineof which, decisions can be made to grant membership to an individual in a group. Similarly, newlineindividuals can decide whether to join the group on the basis of quantified values of trust. The newlinetrust model has been illustrated by assessing the suitability of a set of Group Key Agreement newlineprotocols to applications having different trust requirements. newline