High-efficiency implementation of Toeplitz Strong Extractor for PRNG and QRNG output on CPU/GPU hardware systems

作     者：Anurag, K. S. V. Raghavan, G. Raju, P. Kanaka 

作者机构：Def Inst Adv Technol Sch Quantum Technol Pune 411025 Maharashtra India 

出 版 物：《PHYSICA SCRIPTA》 (Phys Scr)

年 卷 期：2024年第99卷第7期

页      面：075115-075115页

核心收录：

学科分类：07[理学] 0702[理学-物理学] 

基　　金：DIAT(DU) 

主　　题：randomness extraction quantum communication QRNG post-processing Toeplitz Hashing strong extraction 

摘      要：Random Number Generators (RNGs) are devices whose utility spans from cryptography to gambling. Depending on the source of the random seed, RNGs are classified as Pseudo-Random Number Generators (PRNGs), which are based on mathematical algorithms, True Random Number Generators (TRNGs) sourced through seemingly random physical processes, and finally, Quantum Random Number Generators (QRNGs) which harness the intrinsic randomness of measured outcomes of quantum states. To ensure that the output of a QRNG is private and nearly uniform, the raw data undergoes a strong extraction process such as Toeplitz or Trevisan Hashing. The extraction process needs to be extremely efficient to achieve high-speed random number generation. The present work demonstrates state-of-the-art methods for performing information-theoretically provable randomness extraction using variations of the Toeplitz Hashing algorithm. These algorithms are implemented on various hardware for comprehensive analysis. Subsequently, these methods are applied to two raw data sets from a PRNG source and a quantum source. A new benchmark of 19.5 Gbps for Randomness Extraction using the Fast Fourier Transform-based Toeplitz Strong Extractor is demonstrated in this work. The implementation of the algorithm on a GPU-based system demonstrates notably enhanced speed, marking a significant leap beyond the existing state-of-the-art implementations.