Machine Learning assisted excess noise suppression for continuous-variable quantum key distribution

Abstract

Excess noise is a major obstacle to high-performance continuous-variable quantum key distribution (CVQKD), which is mainly derived from the amplitude attenuation and phase fluctuation of quantum signals caused by channel instability. Here, an excess noise suppression scheme based on equalization is proposed. In this scheme, the distorted signals can be corrected through equalization assisted by a neural network and pilot tone, relieving the pressure on the post-processing and eliminating the hardware cost. For a free-space channel with more intense fluctuation, a classification algorithm is added to classify the received variables, and then the distinctive equalization correction for different classes is carried out. The experimental results show that the scheme can suppress the excess noise to a lower level, and has a significant performance improvement. Moreover, the scheme also enables the system to cope with strong turbulence. It breaks the bottleneck of long-distance quantum communication and lays a foundation for the large-scale application of CVQKD.

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.…”

“…A linear activation function is commonly used in the output layer of NNs designed for regression, such that the output can take any continuous value, while sigmoid activation functions are commonly used for classification NNs to return discrete outputs. In the surveyed works, MLPs were applied to excess noise filtering in [12], parameter optimization in [26], reconciliation in [29], and key rate estimation in [30]. Figure 5a gives an example of a generic fully connected MLP, where the connections between the input layer, hidden layer, and output layer are shown, as well as an example perceptron in Figure 5b, indicating how the inputs and weights from the previous layer are transferred into an output via the activation function (following the path outlined in red).…”

“…The value of k should be tuned to minimize misclassification. Turbulence strength classification using a KNN algorithm was undertaken in [12], while DM CV-QKD state classification using KNN algorithms was undertaken by [21,22,24]. Note that no works on ML-assisted CV-QKD utilized an NN for classification.…”

“…The second was the correction of the signal quadratures in post-processing using ∆φ estimates [5,6,15]. The third was, more broadly, the excess noise filtering of the signal quadratures in post-processing [12,14], not focusing specifically on the phase noise. The fourth was correction of the RLO wavefront using reference pulse wavefront estimations [16].…”

“…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.…”