Design And Implementation Of Practical Quantum Key Distribution System

Quantum key distribution(QKD)provides a way for two remote users to securely share a series of random bits while being eavesdropped.Unconditional security can be achieved through the combination of QKD and one-time-pad method.Since the first QKD protocol was proposed by Bennett in 1984,QKD has gone from theoretical analysis and experimental verification to practical application.Reducing the complexity of equipment,improving the efficiency of the system and speeding up the process of practical application are the focus of current research.In this paper,researches are carried out from two aspects: multi-protocol QKD device and phase calibration method for twin-field quantum key distribution(TF-QKD).The specific research contents and results are as follows:1.In order to meet the different requirements of QKD devices in various situations,we propose a QKD structure in this thesis which is compatible with multi-protocols including BB84 protocol,Scarani-Acin-Ribordy-Gisin protocol,reference frame independence(RFI)protocol and six-state protocol.At the same time,the codec is not affected by insertion loss and polarization selectivity of phase modulator.In this paper,the QKD structure and the working principle of the key devices are introduced,and the BB84 protocol and RFI protocol are taken as examples to describe the working principle of the device.Finally,the RFI protocol is taken as an example to carry out experimental verifications.The experimental results coincide with the theoretical simulation,which proves the practicability of the device.2.Aiming at solving the problem of low transmission efficiency caused by phase calibration in the TF-QKD,we propose a phase calibration scheme based on neural network accelerated by field programmable gate array.Phase shift is one of the crucial problems to realize TF-QKD experiments.The current solution is to multiplex the signal light with a reference light and carry out phase calibration with the reference light.However,the transmission efficiency of signal light will be reduced through this way,which can only reach about 50% according to current research.To solve this problem,we develop a time-aware long short-term memory neural network and deployed it onto field programmable gate array after compression,improving the transmission efficiency of TF-QKD to more than 80%,while maintaining a high interference visibility and interference stability.