A Study of Gate-Based and Boson Sampling Quantum Random Number Generation on IBM and Xanadu Quantum Devices

TL;DR

This study evaluates the practicality of quantum random number generation on IBM gate-based and Xanadu boson sampling devices, analyzing output quality and efficiency with post-processing and statistical tests.

Contribution

It compares two quantum platforms for QRNG, implementing debiasing methods and assessing their output quality and efficiency on real hardware.

Findings

Unbiased bitstreams achieved on both platforms

Throughput remains low for quantum devices

High cost per random bit compared to specialized QRNGs

Abstract

Quantum mechanics offers a fundamentally unpredictable entropy source due to the intrinsic probabilistic nature of quantum measurements, making it attractive for secure random number generation. This paper explores the practicality of generating random numbers from two quantum platforms: gate-based circuits on IBM Quantum and (Gaussian) boson sampling with Xanadu Borealis. We implement simple post-processing methods, including the classic Von Neumann extractor and two tailored variants designed to address the correlated structure of boson sampling outputs. We evaluate debiased output from real quantum hardware using the NIST SP800-22r1a test suite and measure the extraction efficiency of each debiasing method. Results show that, while unbiased bitstreams can be achieved on both platforms, throughput remains low and cost per random bit is high compared to specialized QRNG devices.