A robust embedded biometric authentication system based on fingerprint and chaotic encryption

Fingerprint recognition is a reliable solution in user authentication systems. Nevertheless, the security and secrecy of the users data are a concern in today's biometric systems and most of the security attention is focused in biometric template protection to avoid identity theft. In recent years, several approaches have been presented where the main objective is to have the biometric revocable, but almost all them have failed in verify their security and robustness. In last years, chaotic systems have been proposed in cryptography due they have several properties related with cryptography properties such as extreme sensibility on initial conditions with confusion and ergodicity with diffusion. In this paper, we present a novel fingerprint template protection scheme based on chaotic encryption by using the logistic map and Murillo-Escobar's algorithm (Murillo-Escobar et al., 2014). In addition, we present a novel implementation of our scheme in a 32 bit microcontroller for secure authentication systems to show its application on embedded expert systems. In contrast with recent approaches presented in literature, we present a complete security analysis in both statistical and implementation level, to justify the proposed scheme in a real application. Based in the results, the proposed embedded authentication system is secure, effective and at low cost, and it could be implemented on real secure access control systems.