Pseudorandom Functions From Lwe: Rka Security And Application

Pseudorandom Functions (PRF) is a basic primitive in cryptography. In this paper, we study related key attacks (RKA) with which the adversary is able to choose function phi and observe the behavior of the PRF under the modified secret key phi(k). We focus on the PRF from the Learning with Errors (LWE) assumption by Banerjee and Peikert in CRYPTO 2014. We prove that the PRF is secure against unique-input key shift attacks and restricted affine attacks. After that, we use this RKA-secure PRF to construct a robustly reusable fuzzy extractor, which enjoys higher efficiency and better error correction rate.