Phase Compensation for Continuous Variable Quantum Key Distribution Based on Convolutional Neural Network

Abstract

Phase drift extremely limits the secure key rate and secure transmission distance, which is non-negligible in local oscillation continuous variable quantum key distribution (LLO CV-QKD). In order to eliminate the impact caused by phase drift, we analyze the phase noise of the system and propose a phase compensation method based on convolutional neural network (CNN). Moreover, the compensation is performed on the signal according to the estimated value of phase drift before coherent detection. In numerical simulation, we compare the performance of phase compensation methods based on CNN and Kalman filter (KF), and the results show that CNN-based phase compensation has higher accuracy and stability.

A Survey of Machine Learning Assisted Continuous-Variable Quantum Key Distribution

“…Generic CV-QKD protocol, where [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24] were applied during measurement, ref. [28] was applied to key sifting, ref.…”

“…In the context of this review, we note that ML has been used for ∆φ reduction at different stages of the signal quadrature measurement process, including being embedded in the system such that the ∆φ estimate is incorporated into the phase-correction of the RLO [13], the correction of the measured signal quadratures in post-processing using the estimated ∆φ [5,6,15], and the application of an estimated correction to the wavefront of the RLO [16]. Moreover, ML offers potential improvements in computational efficiency for key sifting, reconciliation, and key rate estimation procedures, improving the feasibility of real-time CV-QKD.…”

“…Figure 5d then gives an example convolution of an input array to a feature map. Surveyed works include CNNs developed for wavefront correction in [16] and ∆φ reduction in [13].…”

“…While many traditional approaches have been suggested to overcome the limitations on CV-QKD, machine learning (ML) has recently been shown to have advantages in terms of phase error estimation and excess noise filtering [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20], state discrimination [21][22][23][24], parameter estimation and optimization [25][26][27], key sifting [28], reconciliation [29], and key rate estimation [30,31]. ML-based phase error estimation and noise filtering algorithms offer the potential of improved filtering capabilities due to their ability to map complex relationship between inputs and outputs based on the data alone, without being based on idealistic models that may not represent reality.…”

Machine learning and time-series decomposition for phase extraction and symbol classification in CV-QKD

“…Phase recovery is a critical issue, especially considering the smaller relative rate of information transfer in QKD compared to optical communication systems. [8] utilized a neural network for phase forecasting, which entails a high computational complexity, while [9] used a simpler Bayesian inference-based technique. For symbols demodulation, a simple region-based approach is typically employed [10,11].…”