Daniela Frauchiger

In security applications, it is necessary to use random numbers with the highest possible entropy. Quantum Random Number Generators (QRNG) are most suitable for creating such numbers due to the non-deterministic nature of quantum physics. In this work we present the different steps of the construction of a new design for a high-speed QRNG. Furthermore, in order to make our QRNG suitable for use in security applications, we explain the steps we are taking to make it conform to security standards. Such standards exist and define guidelines for the design of classical Random Number Generators. However, for QRNG, it is necessary to make adaptations, which requires additional work in its design, realization and evaluation with respect to security standards.

A natural definition for a process to be truly random is that its outcome is not predictable from any information available before the process has been started. The advantage of using quantum systems for random number generation compared to classical approaches lies in the fact that the unpredictability of the randomness can be proven based on physical principles. In practice however, due to imperfections of the devices, the resulting raw randomness also depends on classical noise and therefore does not fulfil this definition either. Here we provide a framework for generating almost perfect true randomness using noisy devices by appropriate post-processing (hashing) of the raw randomness. Compared to previous work on random number generators (RNGs), we take this noise into account as side information, which is necessary to meet the above definition of true randomness. Our approach assumes that the process generating the raw randomness is correctly described by a quantum model. Compared to device- independent randomness expansion, this has the advantage of being practical (it is applicable to commercially available devices) and does not require pre-existing randomness. We stress, however, that our results do not rely on any completeness assumption regarding the model or quantum theory. We illustrate our proposal for a Quantum Random Number Generator (QRNG) based on a beam splitter and show how the post-processing procedure used in some of the current devices can be improved.