Unbiased All-Optical Random-Number Generator
Abstract
The generation of random bits is of enormous importance in modern information science. Cryptographic security is based on random numbers which require a physical process for their generation. This is commonly performed by hardware random number generators. These exhibit often a number of problems, namely experimental bias, memory in the system, and other technical subtleties, which reduce the reliability in the entropy estimation. Further, the generated outcome has to be post-processed to "iron out" such spurious effects. Here, we present a purely optical randomness generator, based on the bi-stable output of an optical parametric oscillator. Detector noise plays no role and no further post-processing is required. Upon entering the bi-stable regime, initially the resulting output phase depends on vacuum fluctuations. Later, the phase is rigidly locked and can be well determined versus a pulse train, which is derived from the pump laser. This delivers an ambiguity-free output, which is reliably detected and associated with a binary outcome. The resulting random bit stream resembles a perfect coin toss and passes all relevant randomness measures. The random nature of the generated binary outcome is furthermore confirmed by an analysis of resulting conditional entropies.
Keywords
Cite
@article{arxiv.1712.02254,
title = {Unbiased All-Optical Random-Number Generator},
author = {Tobias Steinle and Johannes N. Greiner and Jörg Wrachtrup and Harald Giessen and Ilja Gerhardt},
journal= {arXiv preprint arXiv:1712.02254},
year = {2017}
}
Comments
10 pages, 4 figures