Innovation in Quantum Computing: Tri-Type QRNG Produces On-Demand Randomness and Record Speeds
Quantum Random Number Generation (QRNG) has seen constant development as a result of the digital world’s increasing reliance on really random numbers, which are necessary for reliable cryptography, precise Monte Carlo simulations, and sophisticated artificial intelligence. Conventional pseudorandom number generators (PRNGs) provide serious security problems because, once their mathematical algorithm and seed are known, they may be predicted. QRNGs generate naturally random numbers by utilizing intrinsic quantum processes like vacuum noise or photon statistics.
The ability of the majority of earlier QRNGs to produce only one particular kind of random number distribution is a major drawback. Different applications, however, call for different distributions: Rayleigh distributed random numbers are frequently used in fluid mechanics and aerodynamics, Gaussian random numbers are required for financial modelling and climate simulations, and uniformly distributed random numbers are essential for machine learning and cryptography. Up until recently, dealing with this diversity of applications necessitated either deploying multiple different hardware systems or converting one random distribution to another, which sacrifices many secure bits and runs the risk of introducing numerical inaccuracies.
The first tri-type QRNG system, which can concurrently measure quantum and process the results to give uniform, Gaussian, and Rayleigh random bits on demand from the same hardware, has now been successfully created and proven by researchers.
Leveraging Quantum Vacuum Noise for Multi-Distribution Output
This flexibility is attained by the tri-type QRNG using dual-quadrature homodyne detection, which measures the quantum vacuum noise. This system measures both quadrature simultaneously rather than just one, as was the case with earlier single-type QRNGs.
All three necessary distributions are naturally based on the raw data from these measurements:
- Gaussian Distribution: Gaussian-distributed raw random numbers are readily obtained by measuring the (Quadrature) quadrature of the vacuum state using homodyne detection.
- Uniform and Rayleigh Distributions: Digital random numbers that are uniform and Rayleigh distributed can be produced by charting the observations in phase space. The amplitude has a Rayleigh distribution, but the phase angle has a uniform distribution.
Importantly, an FPGA board’s electrical postprocessing step handles all of the functionality to switch between the three distribution options, guaranteeing that the high creation rate of the quantum bits is not adversely affected by the distribution choice.
Speed and Accessibility Benchmarks
With the ability to generate more than 60 Gbits/s (Gbps) of raw bits, the experimental device exhibits remarkable speed. The system provides remarkable secure bit rates following the required randomness extraction procedure, which eliminates classical side channel information and guarantees that the randomness is actually quantum:
- Uniform Random Numbers: A secure bit rate of more than 42 Gbps is shown. These consistent random numbers passed the NIST and Dieharder tests, among other stringent statistical analyses. Researchers pointed out that the uniform extraction rate might be increased to over 68.25 Gbps with further optimisation, particularly the use of an improved high-speed, low-noise anti-aliasing filter.
- Gaussian Random Numbers: Secure bit rate extraction of over 14 Gbps. With a more effective extractor and better filter implementation, this rate might be raised to 22 Gbps.
A fiber-coupled Continuous Wave (CW) laser operating at 1550 nm is used in the hardware implementation of this device. The laser’s light is split and interferes with the quantum vacuum state. An AMD ZCU111 Radio Frequency System on Chip (RFSoC) development board is used to digitally process the measurement signals once they are gathered by two homodyne detectors.
Additionally, the technology is already moving towards accessibility in the real world. After real-time testing, the statistically verified numbers are uploaded to the Cisco Cloud and made available to users via the Cisco Quantum Random Number Service. The system continuously creates secure bits on the FPGA board.
The Challenge of Extraction and Statistical Integrity
A robust randomness extractor must be applied to the noisy raw quantum measurement results in order to ensure excellent security. The technique makes use of the well-known Toeplitz extractor, which is based on the universal hashing function and guarantees net randomness extraction even when using a reusable seed, for uniform random numbers.
Nevertheless, producing high-quality Rayleigh and Gaussian numbers has particular post-processing difficulties:
- Gaussian Extraction: A modified recursive approach based on the Wallace method is used by the system. The Most Significant Bits (MSBs) are chosen using an entropy-based truncation technique in order to suppress classical noise, which usually affects the less significant bits (LSBs). There is currently no theoretical evidence of randomness preservation in the procedure, even if the extracted Gaussian data passed the Goodness-of-Fit (GoF) tests. Although it is still a useful and quick option for modelling, simulation, and probabilistic applications that need statistically correct Gaussian distributions, this restricts its use in high-security cryptography scenarios.
- Rayleigh Extraction: Currently, there is no comprehensive quantum randomness extractor for the Rayleigh distribution. In order to denoize the raw Rayleigh bits, researchers are investigating techniques that use classical filters, like the Savitzky-Golay filter.
The tri-type QRNG‘s successful demonstration marks a significant breakthrough since it provides previously unheard-of speed and versatility in quantum randomness generation from a single platform, meeting the many demands of contemporary security and computational applications.
