Data Reconciliation Acceleration For Continuous Variable Quantum Key Distribution

Quantum information technology is a new discipline in the intersection of quantum physics and information science.Its main branch Quantum Key Distribution(QKD)is a combination of quantum mechanics and traditional cryptography.It relies on the physical properties of quantum mechanics: the quantum non-cloning theorem and the Heisenberg uncertainty principle to ensure the confidentiality and security of the key information transmitted by both parties.Two major branches of quantum key distribution technology,continuous variable quantum key distribution(CVQKD)and discrete variable quantum key distribution(DVQKD)have developed rapidly in recent years,the equipment required for the continuous variable quantum key distribution of the coherent state is a general-purpose fiber optic device,which has strong implementation.Therefore,continuous variable quantum key distribution has great potential.Data reconciliation is an important part of the processing of continuous variable quantum key distribution.The acceleration process of data reconciliation is the focus of this paper.First,a PEG(Progressive Edge-Growth)algorithm is used to generate an H matrix of an LDPC code in which a quantum key distribution data reconciliation.The biggest advantage of this method is that it can generate LDPC codes with certainty distribution and determined code rate.The LDPC codes generated by the goodness distribution obtained by density evolution and differential evolution have good decoding performance.The experiment uses a multi-dimensional data reconciliation scheme.The results show that the LDPC code generated by the PEG method has better decoding performance than the LDPC code generated by the Mackay method when the code rate and the code length are the same.The codeword has a lower convergence signalto-noise ratio,a higher transmission rate,and a longer transmission distance.The experimental results show that when the code length is 105 and the code rate is 0.5,the reconciliation efficiency ? of the check matrix generated by the PEG method reaches 93.4%,and the security key of 5.41 kb/s can be extracted,and the transmission distance is 44.5 km.Secondly,in hardware acceleration,the Open CL language is used in the CPU-GPU heterogeneous platform to accelerate the decoding process,The Open CL compatible optimization of the H matrix storage method is implemented,and the CPU-GPU heterogeneous reconciliation system under the MLC/MSD reconciliation scheme is designed.The SW decoding algorithm GPU kernel acceleration implementation method is introduced.By comparing the decoding rate of the CPU-GPU heterogeneous platform and the CPU platform,it is concluded that the decoding rate using the CPU-GPU heterogeneous platform is about 7.5 times that of the CPU platform alone.