Private Information Retrieval applied to Biometric Verification

MASTER Assignment

Private Information retrieval applied to biometric verification

Type : Master M-CS

Period: Sep, 2021 - Apr, 2022

Student : Vries, M.P. de (Martijn, Student M-CS)

Date Final project: Apr 26, 2022

Thesis

Supervisors:

Abstract:

Biometric verification is convenient, since you always have your ’key’ with you, so there is no risk of forgetting or losing it. However, it also poses a security risk, since biometric characteristics are hard to keep secret and can contain privacy sensitive medical information. To keep this secure biometric template protection can be applied in various ways, including feature transformation and biometric cryptosystems. We propose using an oblivious data structure, namely private information retrieval, to retrieve a biometric template and similarity scores in a biometric verification system. The results show that retrieving the template in most cases is too slow for practical use, but retrieving the similarity scores shows promise for small template sizes.