Research On The Error Reconciliation Algorithm In Continuous Variable Quantum Key Distribution

As a new research direction in cryptography, Quantum Key Distribution(QKD) technology, which has been proven unconditionally secure in theory, provides a perfect solution to the problem of key distribution over long distances in One Time Pad encryption algorithm. The error reconciliation module is the key module of post-processing system of QKD, which a?ects the final key rate and transmission distances of the whole system. Compared with the discrete variable(DV) QKD, the continuous variable(CV)QKD, whose future application is promising, has advantages of simple, stable light source and high detection e ciency. Due to the high computational complexity, the error reconciliation module will be the limitation of practical system. Therefore, the research on CV-QKD error reconciliation algorithm is of great significance in terms of improving the performance of the overall CV-QKD system.This dissertation devotes to the research of error reconciliation algorithm in CVQKD, and aims to propose the corresponding optimized schemes, preparing for the further implementation of high-speed post-processing system. The main work of this dissertation is introduced as follows:Accroding to the Sliced Error Correct(SEC) reconciliation algorithm, an exhaustive scheme is proposed, which includes the optimized quantization scheme, the design of estimation function and the reconciliation module. Regarding to the quantization and estimation module, the quantization intevals are obtained iteratively to minimize the information loss. The maximum likelihood criterion is ultilizd to decline the BER between communication parities. The reconciliation scheme is mainly focused on the optimization of the construction of check matrix and decoding module. For the construction of check matrix, the PEG greedy algorithm is selected and the optimized degree distribution is obtained iteratively by the Gaussian approximation density evolution. For the decoding module, through analyzing each steps in inner-level decoding module, the computational complexity of iterative operation, which is the most time consuming, is reduced. So the decoding time of inner-level is decreased evidently. An improved soft information based inter-level decoding scheme is proposed to accelerate the process, which an reduce the iteration rounds greately.Regarding to the multidimensional reconciliation algorithm, a detailed scheme is proposed including the spherize conversion and reconciliation module. The mapping from continuous variable to discrete variable is completed by the spherize conversion method,which avoids the information loss. The reconciliation module is designed through two aspects of binary LDPC and non-binary LDPC, which is mainly focused on. For the non-binary LDPC decoding scheme, the computational complexity of decoding steps,which consume a lot of processing time, is reduced. An improved decoding algorithm in logarithmic domain is proposed after analyzing the steps of non-binary Fast Fourier Transformation Brief Propagation(FFT-BP) decoding algorithm, to avoid the problem of insu cient data accuracy and long processing time, which is brought by the complex multiplication operations.To verify the SEC reconciliation scheme and the multidimensional reconciliation scheme in this dissertation, exhaustive simulations are designed. Compared with other known schemes, The simulation results show our scheme can improve the reconciliation rate under the guarantee of satisfactory reconciliation e ciency.