Garbled Circuit And Homomorphic Encrpytion Based Research Of Methods For Fingerprint Authentication

Biometric data is playing an important role frequently in identity authenticationand identification of activities in business management、government affairs、financeand commerce. Most of all, the privacy of biometric data should be protected in theseapplications mentioned above, especially in open network. Biometric data may has arisk of stealing and impersonation under unsafe transfer. The purposes of this paperare the research of two-party secure function evaluation, and the identityauthentication based on fingerprint minutiae. The main contents would show below:⑴Methods are presented to perform the fingerprint authentication based ondistance metric:①Euclidean distance based method for minutiae matching,②Cityblock distance based method for minutiae matching;⑵Method are presented to perform the fingerprint authentication based onset intersection.In methods based on distance metric, Euclidean distance and City block distanceall are calculated through homomorphic encryption, and then garbled circuit is used tofinish the matching of the corresponding minutiaes; In method based on setintersection, a novel algorithm is presented to perform the minutiae matching. In theproposed algorithm, a sequence consisting of the template of the server and queryminutiae of the user is firstly sorted by oblivious sorting algorithm. Then, all theneighbor elements of the sequence are tested if they are equal. Finally, several addersare used to compute the quantity of matched elements. The proposed algorithm iscrystallized by using garbled circuits. Experiment results and analysis show that,contrast to latter method, the former has a higher accuracy of authentication but morecomplex computation and communication.Aimed to realize the fingerprint authentication between the equipments withlimited resource(mobile or embedded equipments)and the remote sever, someimprovements have been taken in method based on set intersection: the design ofpipeline and tamper-proof token based implementations for fingerprintauthentication. Experiment results and analysis show that the improved algorithm can relieve the communication pressure between the server and the mobileequipments effectively, and be capable of remote fingerprint authentication onmobile or embedded equipments with low memory.