Parallel Real-time Post-processing Of Quantum Random Number Generator

High-speed random number generation is essential in the field of quantum random number generation and applications such as quantum key distribution and quantum communication,high-quality.The quality of random numbers ultimately determines the achievable security and confidentiality.Quantum random number generators exploits inherent uncertainty essence of quantum physics to generate high-quality true random numbers that can be applied to high-security-requirements areas.For the basic characteristics of true random numbers,modern theory has a more comprehensive expression:1)Unpredictability: The elements in the random sequence are independent of each other.Even if the random number generation mechanism is disclosed and a part of the random numbers is known,the remaining elements still cannot be guessed.2)Statistical uniformity: Each element in the random sequence is subject to an equal probability distribution,that is,the probability that each bit(“0” or “1”)appears 50%.3)Non-repeatability: Even if the generating conditions and methods are exactly the same,the same random sequence cannot be generated repeatedly.Among various QRNGs,schemes based on continuous variable quantum fluctuations,including device trusted,device-independent and semi-device-independent schemes,have become promising for its convenience of state preparation,insensitivity of detection efficiency,high measurements bandwidth,and stable and compact optical setup.In such schemes,the model between the ideal random number exploiting process and defections in practical implementation can be established for information-theoretical provable security.However,generic to most QRNGs,who can be classified as serial post-processing,potential limitations exist in these systems.For example,such serial structure of system always faces the rate bottleneck,especially in the step of postprocessing.But,in the fields like secure communications,there are higher requirements for the rate of random number generation.And as for high throughput of true random numbers,system has stringently demands on electronic components and circuits and eventually leads to expensive cost.In recent years,information-theory-provable Toeplitz real-time postprocessing applied to the QRNG is mainly implemented on field programmable gate arrays(FPGA).In the case of taking up a lot of resources of FPGA,it is necessary and universal to consider resource allocation in FPGA development.However,in the previous work of applying FPGA to Toeplitz post-processing,the consideration of the logic resource problem only mentioned the method of splitting the Toeplitz matrix.At present,there has not been specific and quantitative analysis of the allocation of logic resources within the FPGA.In response to the above problems,the main work of this article is as follows:(1)The vacuum fluctuation is selected as the entropy source of the quantum random number generator,and three frequency bands of the entropy source are selected as three channels to construct a parallel structure of the random number generator.(2)Parallel Toeplitz hash post-processing corresponding to multiple quantum frequency bands in field programmable gate array is implemented.The logic resource occupancy of post-processing with different Toeplitz operation channel numbers is analyzed,and the specific process and signal timing of Toeplitz post-processing are designed to fully utilize the advantages of parallel computing and the hardware resources of field programmable gate arrays.(3)The quality of the random numbers generated by the random number generator are tested and verified.This work ultimately achieved a real-time post-processing rate higher than 8 Gbits/s.The quantum random numbers generated by each channel can pass the NIST test.The autocorrelation and cross-correlation coefficients of the random numbers remain a low level.And the whole system implementation can maintain high random number generation quality for at least 10 hours.Compared with the previous work,the work of multi-channel parallel real-time post-processing based on the specific situation of FPGA described in this article can greatly improve the resource utilization rate of FPGA.In the case of the same cost(using the same FPGA chip),the rate of random number generation by the final random number generator can be greatly increased.If an ADC with a higher sampling rate of 1GHz and an FPGA with corresponding performance,the real-time random number generation rate is expected to exceed 34 Gbps,but the current experiment is limited by cost.