Fuzzy Identity-Based Signature Scheme Suitable for Biometric Authentication

The security of a signature scheme given in the standard model (SM) will be more stable and reliable than that given in the random oracle model (ROM). Fuzzy identity-based signature (FIBS) enables a user to generate a signature for a set of descriptive attributes, defined as ω=ωjj=1n. Any attributes set ω'=ωj'j=1n can validate the signature provided that the distance between ω and ω' is below a predefined threshold. Most of the existing FIBS schemes are based on the ROM. It is of great significance to design a FIBS scheme based on the SM. In this work, we adopt fingerprint minutiae as the biometric modality and present a feature extraction algorithm E that transforms raw minutiae into quantized, privacy-preserving attribute sets, and we present a False Rejection Rate (FRR)-False Acceptance Rate (FAR) trade-off framework to calibrate matching threshold t, with adjustable n for qualified error performance. Subsequently, we present a novel and efficient FIBS scheme, which is proven to be unforgeable in SM for any polynomially bounded adversary under selective identity attack model. Compared to the existing FIBS schemes based on the ROM, our new FIBS scheme has a strong security model. Compared to the existing FIBS scheme based on the SM, our new FIBS scheme reduces total computation consumption by approximately 33.55% and achieves a significant reduction in communication consumption, saving approximately 68.07% of the message and signature size, which is suitable for biometric authentication.