Research On The Optimization Of The Plug-Play Reference-Frame-Independent Measurement-Device-Independent Quantum Key Distribution System

Traditional information encryption technology mostly relies on the assumption of difficult mathematical problems,but the development of quantum computing takes a challenge to this.Quantum key distribution relies on the basic principles of quantum mechanics to realize unconditional security,which has become an important means to solve the problem of information transmission.After decades of research,quantum key distribution technology has become practical.The plug and play reference-frame-independent measurement-device-independent quantum key distribution(Plug-Play RFI-MDI-QKD)system combines plug-and-play structure,reference frame independent design,and measurement device independent protocol,which solves the problem of the quantum state being affected by the channel and drift of the system reference of the key distribution parties,and greatly reduce the equipment complexity of the QKD system.The key rate of the QKD system is closely related to the parameters selected during system operation.Choosing the optimal operating parameters can increase the secure key rate of the QKD system.Therefore,the parameter adaptive optimization of the QKD system is very important.The traditional optimization algorithm takes a long time to search for the optimal parameters,which makes the system control untimely and causes the performance of the QKD system to decrease.The Plug-Play RFI-MDI-QKD system uses three bases and the plug-play structure brings untrusted source problem.Compared with other QKD systems,there are more parameters to be optimized,making the optimization algorithm more time-consuming.This thesis mainly focuses on the parameters optimization of Plug-Play RFI-MDI-QKD system.The main research contents and results are as follows:Firstly,jointly optimize the 9 parameters of the Plug-Play RFI-MDI-QKD system under different conditions,including the internal transmittance of the signal state_??,the internal transmittance of the decoy state_??,the probability to choose signal state P_?,the probability to choose decoy state P_?,the conditional probability to choose Z basis conditional signal stateP_Z|?,the conditional probability to choose X basis conditional signal stateP_X|?,the conditional probability to choose Z basis conditional decoy stateP_Z|?,the conditional probability to choose X basis conditional decoy stateP_X|?and the fluctuation range of incident photon numbers?,and use coordinate descent algorithm as the parameter optimization algorithm.Numerical analysis results show that the key rate of full parameter optimization and partial parameter optimization is higher than that of the unoptimized system key rate,and the system key rate of the full parameter optimization is two orders of magnitude higher than that of the unoptimized system,and the key rate is equivalent to the Non-Plug-Play RFI-MDI-QKD system;The parameter optimization under different key length is studied,and the results show that the system key rate increases with the increase of the key length;The parameter optimization under different light source intensities is studied,the results show that a stronger light source can improve the key rate and security key distance of the system.Secondly,a Plug-Play RFI-MDI-QKD system based on neural network is proposed.Construct 9 neural networks and use the Levenberg-Marquardt algorithm to train the neural networks.The trained neural network is used to predict the optimal parameters of the system.Numerical analysis results show that the accuracy of the optimal parameters predicted by the neural network under different conditions is more than 99%,and the calculation time is reduced by two orders of magnitude compared with traditional parameter optimization methods.The effectiveness and feasibility of the neural network for optimizing the parameters of the Plug-Play RFI-MDI-QKD system are verified;the coding control module of the system is designed and implemented based on the Field Programmable Gate Array(FPGA)platform,and the synchronization signal detection module and coding selection module are designed and implemented;The neural network module is designed and implemented.The test results show that the average output error of the neural network module is less than 1%,which can be used to predict the optimal parameters of the system.