High-Secure Fingerprint Authentication System Using Ring-LWE Cryptography

This paper presents a high-secure fingerprint authentication system using ring learning with errors (ring-LWE) cryptography to protect users' fingerprint data more securely. A delay-optimized high-accuracy scheme for a fingerprint-features extraction approach is proposed to collect necessary features' information from fingerprint images. In addition, a ring-LWE cryptography scheme using low-latency number theoretic transform (NTT) polynomial multiplications is deployed to speed up the ring-LWE encryption and decryption times. As a result, the processing time of the fingerprint authentication system is significantly reduced, and the fingerprint data are effectively protected. The simulation results show that the proposed NTT multiplication-based ring-LWE cryptography scheme for fingerprint features outperforms the existing works up to 46% and 44% in terms of encryption time and decryption time, respectively. The latency of the whole fingerprint authentication system is less than 160 ms, which makes it suitable for practical applications. Furthermore, performance analysis on entropy and similarity of the encrypted fingerprint features proves the domination of the proposed system compared with the previous systems in terms of confidentiality.