Cryptographically-Secure and Efficient Remote Cancelable Biometrics Based on Public-Key Homomorphic Encryption

Cancelable biometrics is known as a template protection approach, and concrete protocols with high accuracy and efficiency have been proposed. Nevertheless, most known protocols, including the Hattori et al. protocol (Journal of Information Processing, 2012), pay little attention to security against the replay attack, which leads to severe authenticity violation in the remote authentication setting. In this paper, we revisit the Hattori et al. protocol based on the Boneh-Goh-Nissim encryption scheme, and propose a secure variant while keeping user-friendliness of the original protocol. Our protocol uses the revocation method of the original protocol in a proactive manner, i.e., in our protocol, the public key assigned to a user is randomly re-generated in every authentication process. We define a general and formal security game that covers the replay attack and considers fuzziness of biometric feature extraction, and show that our protocol is secure in that model. The computation and communication costs of our protocol are more efficient than those of similar protocols.