Enhanced Template Protection Algorithms Based on Fuzzy Vault and Cuckoo Hashing for Fingerprint Biometrics

Abstract

Biometrics provides better authentication. Unprotected biometrics is open to attacks from intruders as stolen biometrics may not be revocable. Although there are several points where attacks can be launched on biometric systems, template databases are said to be the most frequently attacked. When a template database is attacked, attackers can add fresh templates, modify the existing ones, copy or steal templates and later construct a spoof from it or replay it back into the biometric system to impersonate a genuine user. Several template security systems have been presented in the literature to secure biometric templates. Fuzzy vault, as proposed by many researchers is, to some extent, one of the best algorithms to achieve template protection as it has good security. Fuzzy vault, however, lacks irreversibility, revocability, and diversity. To address these disadvantages and strengthen fuzzy vault, this study combines a noninvertible feature transformation template protection algorithm known as cuckoo hashing that possesses irreversibility, revocability, and diversity properties with a fuzzy vault for privacy. The study used fingerprint biometrics as it is widely used. The proposed algorithm was implemented in the MATLAB 2016a environment using FVC 2004 DB1 fingerprint public database. The proposed algorithm recorded a FAR of 0.01% and an FRR value of 0.09% with an EER of 0.05%.