Abstract: An emerging approach to the problem of identity theft is represented by the adoption of biometric authentication systems. Such systems however present several challenges, related to privacy, reliability and security of the biometric data. Inter-operability is also required among the devices used for authentication. Moreover, very often biometric authentication in itself is not sufficient as a conclusive proof of identity and has to be complemented with multiple other proofs of identity such as passwords, SSN, or other user identifiers. Multi-factor authentication mechanisms are thus required to enforce strong authentication based on the biometric and identifiers of other nature. In this paper we propose a two-phase authentication mechanism for federated identity management systems. The first phase consists of a two-factor biometric authentication based on zero knowledge proofs. We employ techniques from the vector-space model to generate cryptographic biometric keys. These keys are kept secret, thus preserving the confidentiality of the biometric data, and at the same time exploit the advantages of biometric authentication. The second phase combines several authentication factors in conjunction with the biometric to provide a strong authentication. A key advantage of our approach is that any unanticipated combination of factors can be used. Such authentication system leverages the information of the user that are available from the federated identity management system.
