A Statistical Inference Attack on Privacy-Preserving Biometric Identification Scheme

Biometric identification allows people to be identified by their unique physical characteristics. Among such schemes, fingerprinting is well-known for biometric identification. Many studies related fingerprint-based have been proposed; however, they are based purely on heavy cryptographic primitives as additively homomorphic encryption and oblivious transfer. Therefore, it difficult apply them large databases because of the expense. To resolve this problem, some schemes proposed that simple matrix operations rather than primitives. Recently, Liu et al. an improved matrix-based scheme using properties orthogonal matrices. Despite being more efficient when compared previous systems, still fails provide sufficient security against various types attackers. In paper, we demonstrate vulnerable attacker who operates with a cloud server introducing statistical-inference attack algorithms. Moreover, propose concrete identity confirmation parameters adversary must always pass, present experimental results our algorithms both feasible practical.