International Association for Cryptologic Research

International Association
for Cryptologic Research

CryptoDB

A Refined Hardness Estimation of LWE in Two-step Mode

Authors:
Wenwen Xia , Xidian University
Leizhang Wang , Xidian University
Geng Wang , Shanghai Jiao Tong University
Dawu Gu , Shanghai Jiao Tong University, Xidian University
Baocang Wang , Xidian University
Download:
Search ePrint
Search Google
Presentation: Slides
Conference: PKC 2024
Abstract: Recently, researchers have proposed many LWE estimators, such as lattice-estimator (Albrecht et al, Asiacrypt 2017) and leaky-LWE-Estimator (Dachman-Soled et al, Crypto 2020), while the latter has already been used in estimating the security level of Kyber and Dilithium using only BKZ. However, we prove in this paper that solving LWE by combining a lattice reduction step (by LLL or BKZ) and a target vector searching step (by enumeration or sieving), which we call a Two-step mode, is more efficient than using only BKZ. Moreover, we give a refined LWE estimator in Two-step mode by analyzing the relationship between the probability distribution of the target vector and the solving success rate in a Two-step mode LWE solving algorithm. While the latest Two-step estimator for LWE, which is the “primal-bdd” mode in lattice-estimator1, does not take into account some up-to-date results and lacks a thorough theoretical analysis. Under the same gate-count model, our estimation for NIST PQC standards drops by 2.1∼3.4 bits (2.2∼4.6 bits while considering more flexible blocksize and jump strategy) compared with leaky-LWE-Estimator. Furthermore, we also give a conservative estimation for LWE from the Two-step solving algorithm. Compared with the Core-SVP model, which is used in previous conservative estimations, our estimation relies on weaker assumptions and outputs higher evaluation results than the Core-SVP model. For NIST PQC standards, our conservative estimation is 4.17∼8.11 bits higher than the Core-SVP estimation. Hence our estimator can give a closer estimation for both upper bound and lower bound of LWE hardness.
BibTeX
@inproceedings{pkc-2024-34035,
  title={A Refined Hardness Estimation of LWE in Two-step Mode},
  publisher={Springer-Verlag},
  author={Wenwen Xia and Leizhang Wang and Geng Wang and Dawu Gu and Baocang Wang},
  year=2024
}