100Gbits! Quantum random number generation sets another record

A team of physicists from Ghent University - Intercollegiate Centre for Microelectronics, Technical University of Denmark and Bari University of Technology recently reported that random numbers can be generated faster than standard methods using quantum fluctuation (Quantum fluctuation). In their study, reported in the journal PRX Quantum, the team uses the behavior of particle and antiparticle pairs to create random generators that are 200 times faster than conventional systems.

Random number generation is important in computer science and is used to create encryption keys for many sensitive applications, in addition to applications such as generating random backgrounds and scenes in video games. Also, random numbers are important for statistical analysis, probability theory and modern computer simulations, digital cryptography and cryptocurrency wallets.

Emerging communication and cryptography applications require reliable, fast and unpredictable random number generators. Due to the randomness inherent in quantum mechanics, quantum random number generation allows the generation of truly unpredictable numbers. One popular approach is to use quantum vacuum states to generate random numbers. Although convenient, the speed of this method is generally limited compared to other schemes.

In this study, with a custom co-design of optoelectronic integrated circuits and digital filtering to reduce side information, this experiment demonstrates an ultra-fast generation rate of 100 Gbit/s, setting a new record for vacuum-based quantum random number generation by an order of magnitude. In addition, the experimental demonstration is well supported by an upgraded device-dependent framework that is secure for both classical and quantum side information and also correctly accounts for nonlinearities in the digitization process. This ultrafast secure random number generator on a chip-scale platform offers promise for next-generation communication and cryptography applications.

The Quantum Random Number Generator (QRNG) generates random numbers by measuring an arbitrary product Q of vacuum states |0> using zero-difference detection. In practice, a balanced zero-difference detector, consisting of an optical hybrid element, a pair of balanced photodiodes, and a transimpedance amplifier, is used whose output is digitized using an analog-to-digital converter (ADC) and further distilled into a sequence of truly random bits.

Fig. 1 Block diagram of a quantum random number generator based on quadrature measurements of the vacuum state |0>.

In general, the speed of vacuum noise-based RNGs is limited by the speed of the balanced zero-difference detector and its noise performance. By integrating the zero-difference detector, a significant improvement in the emission noise-limited bandwidth is achieved and a higher generation rate is realized.

The practical implementation of the QRNG is shown in Figure 2. A 1550 nm continuous-wave laser (Koheras Basik E15) is fed into a photonic integrated circuit (PIC) fabricated using imec's iSiPP50G silicon photonics platform. the PIC contains a tunable 2 × 2 hybrid element connected to two photodiodes. The photocurrent is converted to voltage by 100 nm GaAs pseudomorphic high electron mobile transistor (pHEMT) technology and amplified by a linear broadband amplifier (SHF 807) to optimally fill the ADC range.

Next, the analog signal is digitized by a Keysight DSOZ632A digital storage oscilloscope (DSO), which has an internal 8-bit ADC with a sampling rate of 20 GS/s. The captured data is processed offline to balance the detector response and generate a random bit stream based on the proposed minimum entropy framework. In addition to the significant size reduction, the use of a custom integrated balanced zero-difference detector offers the possibility to design the circuit to best fit the application at hand, thus greatly improving the performance compared to discrete off-the-shelf implementations.

In addition, the frequency response of the detector shows a gradual gain drop at high frequencies, which limits the number of FIR taps required to balance the detector response. Also using the minimum entropy framework of Sec.

Figure 2 provides an overview of the QRNG (Quantum Random Number Generator) setup. (a) The vacuum noise used to generate random numbers. (b) Micrographs of a fabricated PIC and TIA. (c) The digitized Gaussian distribution. ( d ) The distribution of random 32-bit integers extracted and grouped into 256 bins.

The metrics needed to estimate the generation rate of the setup shown in Figure 2 can now be measured. The various metrics that should be obtained are the maximum DNL of the ADC, the PSD and excess noise of the zero-difference measurement, and finally the FIR-filter tap coefficients.

Figure 3 (a) Heat map of the errors present in the captured sine wave. The inset shows the distribution of codes -50, 0 and 100. The blue line represents INL, the average value of the error for each digitized result. (b) Integral nonlinearity versus ADC coding. (c) Differential nonlinearity versus ADC coding.

To verify the improvement in temporal correlation, the team fitted one set of measurements and applied it to another independent data set. When examining the autocorrelation of the second set of data before and after equalization, as shown in the figure below, it is clear that there is indeed a significant improvement in the temporal correlation after equalization.

Figure 4 Effect of equalizer on autocorrelation. The autocorrelation coefficient is the average of 10,000 measured data sets.

In this work, an integrated quantum random number generator achieves a generation rate of 100 Gbit/s based on vacuum rise and fall. This generation rate is obtained by applying a framework that is secure to both classical and quantum side information. A method to measure static ADC nonlinearity is established. The limited bandwidth present in the receiver is then compensated for by applying detector equalization, reducing the amount of loss due to quantum-side information leakage. The rate of 100 Gbit/s is much faster compared to recent vacuum fluctuation based random generators.

Previously, one of the limiting factors for achieving high generation rates using vacuum fluctuations was the presence of classical noise in the measurements. By using custom application-specific integrated circuits, the team demonstrated that this bottleneck can be greatly reduced, proving that vacuum fluctuation-based quantum random number generators (QRNG) are a viable solution for applications requiring high generation rates.

Reference links:

[1] https://journals.aps.org/prxquantum/abstract/10.1103/PRXQuantum.4.010330

[2] https://phys.org/news/2023-04-quantum-fluctuations-generate-random-faster.html