LEARNING PARITY WITH NOISE-BASED RELATIONAL ENCRYPTION FOR PROXIMITY RELATIONS

摘要

A method includes receiving a first and a second linearity ciphertexts representative of a first and second biometric templates, respectively that are encrypted using a relational linearity encryption scheme (linearity scheme). The linearity scheme is based on learning parity with noise. The method includes discovering a linearity relationship between the first and the second linearity ciphertexts. The method includes receiving a first and a second proximity hash value representative of the first and second biometric templates, respectively encrypted using a relational proximity hash scheme (proximity scheme). The proximity scheme is based on the linearity scheme and an error correcting code. The method includes detecting a proximity between the first and the second proximity hash value in terms of a Hamming distance. The method includes authenticating an identity of a user based on the proximity and the linearity relationship.