A cancelable biometric identification scheme based on bloom filter and format-preserving encryption

Abstract

Biometric based authentication systems are being prominently used everywhere. The biometric data, popularly known as a biometric template, is generally stored on the database server in its unprotected form. Unlike passwords, once compromised, biometric data can never be recovered. An ideal biometric system should provide accessibility, acceptability, availability, high security, and high biometric performance to the user. Current deployments generally relax in one or more requirements, resulting in lingering concerns about the privacy and security of individuals’ biometric data. Our proposed work introduces a cancelable biometric template protection scheme based on the format-preserving encryption and Bloom filters. The format-preserving encryption encrypts the biometric template, which then maps to the Bloom filter based template that represents the cancelable template. The use of format-preserving encryption along with Bloom filters helps to achieve the security of the input biometric template and identification with good recognition performance. We achieve 0.2% FRR at 0.01% FAR for IITD-CASIA virtual dataset in the uni-biometric scenario. A comparison with the existing schemes shows that our proposed scheme exhibits high recognition performance for both uni-biometric and multi-biometric datasets, while simultaneously, the security of the overall system is preserved.