Quantum Random Number Generation

last updated 2026-08-31

Physics / mechanism

Quantum random number generation (QRNG) extracts bits from a physical process whose outcomes are fundamentally indeterminate under quantum mechanics, rather than from a deterministic algorithm. The practical task is to isolate a quantum entropy source from classical noise, digitise it, and then apply a randomness extractor that compresses the raw data down to a length bounded by a rigorously estimated min-entropy. Randomness is needed for simulation, cryptography and secure quantum communication.

Several distinct entropy sources are in active use. Laser phase noise exploits spontaneous-emission-driven phase diffusion in a semiconductor laser, converted to intensity fluctuations by an interferometer; it is attractive for its high generation rate and its suitability for photonic integration. Continuous-variable (CV) QRNG performs homodyne measurement on the vacuum state, using an existing payload laser as local oscillator and a digitising ADC. Spatial-mode schemes read out the spatial intensity fluctuations of coherent light across a camera sensor, treating shot noise resolved across many pixels as a parallel entropy source.

Key design parameters are the bandwidth of the entropy source, the probability distribution of the raw samples, the ADC resolution, and the extractable randomness per sample. For laser-phase-noise QRNG a full physical model predicting the entropy-source power spectrum and the raw-data probability distribution allows the source bandwidth and extractable randomness, and therefore the achievable generation rate, to be estimated and optimised quantitatively. In CV-QRNG the ADC resolution directly sets the yield: a 12-bit ADC applied to a raw record of about 1 Mb produced roughly 19.5 Kb of certified random numbers after formal min-entropy bounding.

A separate branch is device-independent QRNG, in which randomness is certified from observed measurement statistics rather than from a trusted device model. This requires efficient detection of quantum states, and high-dimensional encoding is a route to improved noise resilience and information capacity; the experimental bottleneck is the loss and polarisation sensitivity of the active modulators normally needed for basis selection.

Competitive landscape

ApproachEntropy sourceReported figuresNoted constraint
Laser phase noiseSpontaneous-emission phase diffusionRate-optimisation model validated; high rate claimedComplete theoretical model for optimal rate was previously incomplete
CV homodyne (vacuum)Vacuum-state quadrature fluctuations~1 Mb raw per satellite pass, ~19.5 Kb certified, 12-bit ADCYield limited by raw key length and ADC resolution
Spatial quantum noise (EMCCD)Spatial intensity fluctuations of coherent states5.92 Gbps instantaneous without algorithmic extraction; 7.5 Mbps sustainedSustained rate capped by serial electronic readout bandwidth
Device-independent / high-dimensionalCertified from measurement statisticsPoled-fibre phase modulator proposed to replace lossy active modulatorsModulator loss and polarisation sensitivity in high dimensions

The trade-off across these approaches is between raw rate, integrability, and the strength of the security assumption. Trusted-device schemes (phase noise, CV homodyne, spatial noise) deliver higher throughput but depend on a validated device model for the min-entropy bound; device-independent QRNG weakens those assumptions at the cost of detection efficiency and modulator overhead.

Evidence base

Frontier (open questions)

Synthesised 2026-08-31 from 4 KB sources by the resynth pipeline; citations are KB source slugs.

Frontier questions