| Title |
A Statistical Verification Framework for Gaussian Error Samplers in Post-Quantum Cryptography Hardware Implementations |
| Authors |
정도영(Doyoung Chung) ; 김문석(Moon-Seok Kim) |
| DOI |
https://doi.org/10.5573/ieie.2026.63.9.25 |
| Keywords |
Post-quantum cryptography; Learning with errors; Gaussian error sampler; Quantile-quantile plot; Kullback-Leibler divergence |
| Abstract |
As quantum computing technology advances, concerns regarding the security of conventional public-key cryptographic systems have increased, highlighting the importance of post-quantum cryptography (PQC). In particular, Learning With Errors (LWE)-based cryptographic schemes require error samples that follow a Gaussian distribution to ensure cryptographic security. Recently, hardware-based Gaussian error generation techniques using semiconductor devices, memories, and physical noise sources have been studied to reduce the computational and memory overhead of software-based Gaussian samplers. However, hardware-generated error samples can be affected by process variations, temperature changes, measurement noise, and other non-ideal factors. Therefore, it is necessary to systematically verify their conformity to the target Gaussian distribution. In this paper, we compare the roles and limitations of Q-Q plots, Kullback-Leibler (KL) divergence, Shapiro-Wilk test, and Kolmogorov-Smirnov test for validating Gaussian error samples. Based on this comparison, we propose a multi-metric statistical verification framework that combines these evaluation methods in a stepwise manner. In addition, we present Pass/Warning/Fail decision criteria based on KL divergence, Shapiro-Wilk test pass rate, and Kolmogorov-Smirnov test pass rate, enabling practical suitability assessment of Gaussian error samplers for PQC hardware implementations. |