Quantum-noise based true random number generation

Quantum Random Number Generators (QRNGs) promise information-theoretic security by exploring the intrinsic probabilistic properties of quantum mechanics. In practice, their security frequently relies on a number of assumptions over physical devices. In this thesis, a randomness generation framework...

Full description

Bibliographic Details
Main Author: Ferreira, Maurício de Jesus (author)
Format: masterThesis
Language:eng
Published: 2022
Subjects:
Online Access:http://hdl.handle.net/10773/33737
Country:Portugal
Oai:oai:ria.ua.pt:10773/33737
Description
Summary:Quantum Random Number Generators (QRNGs) promise information-theoretic security by exploring the intrinsic probabilistic properties of quantum mechanics. In practice, their security frequently relies on a number of assumptions over physical devices. In this thesis, a randomness generation framework that explores the amplitude quadrature fluctuations of a vacuum state was analyzed. It employs a homodyne measurement scheme, which can be implemented with low-cost components, and shows potential for high performance with remarkable stability. A mathematical description of all necessary stages was provided as security proof, considering the quantization noise introduced by the analog-to-digital converter. The impact of experimental limitations, such as the digitizer resolution or the presence of excess noise due to an unbalanced detection, was characterized. Moreover, we propose a framework to estimate the excess entropy introduced by an unbalanced detection, and its high impact within the Shannon entropy model was experimentally verified. Furthermore, a real-time dedicated QRNG scheme was implemented and validated. The variance characterization curve of the homodyne detector was measured, and the quantum fluctuations were determined to be preponderant for an impinging power PLO < 45.7mW. By estimating the worst-case min-entropy conditioned on the electronic noise, approximately 8.39 true random bits can be extracted from each sample, yielding a maximum generation rate of 8.23 Gbps. With a lengthcompatible Toeplitz-hashing algorithm, these can be extracted at 75 Mbps with an upper security bound of 2−105, which illustrates the quality of this implementation. Moreover, the generation scheme was validated and verified to pass all the statistical tests of the NIST, DieHarder, and TestU01’s SmallCrush batteries, as well as most of TestU01’s Crush evaluations. Finally, we propose a framework for time-interleaving the entropy source within a classical communication channel, which removes the need for a dedicated generation device. After assessing the conditions where quantum noise is dominant, support for generation rates up to 1.3 Gbps was observed. The random bitstream was subjected to the NIST randomness test suite and consistently passed all evaluations. Moreover, a clean quadrature phase shift keying constellation was recovered, which supports the multi-purpose function of the scheme.