Biometric cryptosystems: authentication, encryption and signature for biometric identities

Zusammenfassung) Biometrics have been used for secure identification and authentication for more than two decades since biometric data is unique, non-transferable, unforgettable, and always with us. Recently, biometrics has pervaded other aspects of security applications that can be listed under the topic of “Biometric Cryptosystems”. Although the security of some of these systems is questionable when they are utilized alone, integration with other technologies such as digital signatures or Identity Based Encryption (IBE) schemes results in cryptographically secure applications of biometrics. It is exactly this field of biometric cryptosystems that we focused in this thesis. In particular, our goal is to design cryptographic protocols for biometrics in the framework of a realistic security model with a security reduction. Our protocols are designed for biometric based encryption, signature and remote authentication. We first analyze the recently introduced biometric remote authentication schemes designed according to the security model of Bringer et al.. In this model, we show that one can improve the database storage cost significantly by designing a new architecture, which is a two-factor authentication protocol. This construction is also secure against the new attacks we present, which disprove the claimed security of remote authentication schemes, in particular the ones requiring a secure sketch. Thus, we introduce a new notion called “Weak-identity Privacy” and propose a new construction by combining cancelable biometrics and distributed remote authentication in order to obtain a highly secure biometric authentication system. We continue our research on biometric remote authentication by analyzing the security issues of multi-factor biometric authentication (MFBA). We formally describe the security model for MFBA that captures simultaneous attacks against these systems and define the notion of user privacy, where the goal of the adversary is to impersonate a client to the server. We design a new protocol by combining bipartite biotokens, homomorphic encryption and zero-knowledge proofs and provide a security reduction to achieve user privacy. The main difference of this MFBA protocol is that the server-side computations are performed in the encrypted domain but without requiring a decryption key for the authentication decision of the server. Thus, leakage of the secret key of any system component does not affect the security of the scheme as opposed to the current biometric systems involving crypto-