Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2507.23117 · Quantum Science and Technology

Neural network for excess noise estimation in continuous-variable quantum key distribution under composable finite-size security

Lucas Queiroz Galvão, Davi Juvêncio Gomes de Sousa, Micael A. Dias, Nelson Alves Ferreira Neto

Published Feb 5, 2026• 2 citations
arXivHTMLPDF

Abstract

Abstract Parameter estimation is a critical step in continuous-variable quantum key distribution (CV-QKD), as the statistical uncertainty from a finite data size leads to pessimistic worst-case bounds that drastically reduce the secret key rate and range. While machine learning techniques have been proposed for this task, they have lacked the rigorous statistical framework necessary for integration into a composable security proof. In this work, we bridge this gap by introducing a statistically rigorous framework for using neural networks for parameter estimation in CV-QKD with quantifiable composable security. We develop a neural network estimator for the excess noise and, crucially, derive its worst-case confidence interval using a delta method approach, ensuring the estimation fails with a probability not exceeding ε PE . This allows the network to be integrated into a parameter estimation protocol that is operationally equivalent to the standard maximum likelihood method but yields significantly tighter parameter bounds. Our numerical results demonstrate that this method provides substantially more precise estimates, which directly translates into a higher secret key rate and extended transmission distance over a fiber channel under a collective Gaussian attack. This work establishes that machine learning can be securely and effectively harnessed to overcome a key performance limitation in practical CV-QKD systems.

DOI
10.1088/2058-9565/ae4268

Related papers

Neural network-based finite-sample transmission coefficient estimation and uncertainty quantification in satellite-to-ground CV-QKD

Jiayan Shao, Lanjian Chen, Mingshan Dong +4 more

Abstract Satellite-to-ground continuous variable quantum key distribution (CV-QKD) is constrained by rapidly varying free-space channels and short transmission windows, which limit the amount of data available for parameter estimation within each channel interval. Under finite-sample conditions, parameter point estimates are subject to statistical uncertainty and may even yield an overly favorable description of the true channel state, while errors in the estimated transmission coefficient directly affect the covariance matrix used in the security analysis. In this work, we investigate the finite-sample effects of transmission coefficient estimation in dynamic free-space CV-QKD. A neural network is introduced to estimate the transmission coefficient, and the Delta method is used to quantify the uncertainty in its predictions. The proposed method is compared with the standard maximum likelihood estimation method. Confidence intervals for the transmission coefficient are constructed for different sample sizes, and the effects of the statistical uncertainties on the secret key rate are analysed. Under the considered Gaussian-channel simulation conditions, the neural network can provide a more favorable nominal lower bound on the transmission coefficient than MLE in certain sample-limited regimes, leading to a higher parameter-estimation-limited key-rate metric. This work establishes a statistical framework that connects neural network-based transmission coefficient estimation and predictive uncertainty quantification with covariance matrix construction and secret key rate analysis. It provides a basis for machine learning-assisted parameter estimation in free-space CV-QKD systems operating with finite samples.

New Journal of Physics•Nov 9, 2026•0 citations

Machine learning for secure key rate in continuous-variable quantum key distribution.

Min-Gang Zhou, Zhiping Liu, Wen‐Bo Liu +6 more

Continuous-variable quantum key distribution (CV-QKD) with discrete modulation has received widespread attentions because of its experimental simplicity, lower-cost implementation and ease to multiplex with classical optical communication. Recently, some inspiring numerical methods have been applied to analyse the security of discrete-modulated CV-QKD against collective attacks, which promises to obtain considerable key rate over one hundred kilometers of fiber distance. However, numerical methods require up to ten minutes to calculate a secure key rate one time using a high-performance personal computer, which means that extracting the real-time secure key rate is impossible for discrete-modulated CV-QKD system. Here, we present a neural network model to quickly predict the secure key rate of homodyne detection discrete-modulated CV-QKD with good accuracy based on experimental parameters and experimental results. With the excess noise of about $0.01$, the speed of our method is improved by about seven orders of magnitude compared to that of the conventional numerical method. Our method can be extended to quickly solve complex security key rate calculation of a variety of other unstructured quantum key distribution protocols.

•Aug 5, 2021•0 citations

Continuous-variable quantum key distribution with noisy squeezed states

Akash nag Oruganti, Ivan D. Derkach, Radim Filip +1 more

Abstract We address the crucial role of noisy squeezing in security and performance of continuous-variable (CV) quantum key distribution (QKD) protocols. Squeezing has long been recognized for its numerous advantages in CV QKD, such as enhanced robustness against channel noise and loss, and improved secret key rates. However, the excess noise of the squeezed states, that unavoidably originates already from optical loss as well as other imperfections in the source, raises concerns about its potential exploitation by an eavesdropper. For the widely adopted trust assumption on the excess noise in the signal states, we confirm the stability of the protocol against the noisy squeezing in both purely attenuating as well as noisy channels in the asymptotic limit, which implies perfect parameter estimation. In the finite-size regime we show that this stability largely holds at up to 107 data points using optimal biased homodyne detection for key distribution and parameter estimation. Untrusted assumption on the noisy squeezing, on the other hand, introduces additional security bounds on the squeezing excess noise already in the asymptotic regime, which is further enforced by the finite-size effects. Additionally, we show the critical negative role of noisy squeezing in the case of atmospheric free-space channels, already in the trusted-noise and asymptotic assumptions, which emphasizes the importance of squeezing purity in the free-space quantum channels. Our results pave the way towards practical realization of squeezed-state CV QKD protocols in both fibre and free-space channels.

arXiv:2404.05247•Quantum Science and Technology•Jan 17, 2025•19 citations

Continuous-variable quantum key distribution with noisy squeezed states

Akash nag Oruganti, Ivan D. Derkach, Radim Filip +1 more

Abstract We address the crucial role of noisy squeezing in security and performance of continuous-variable (CV) quantum key distribution (QKD) protocols. Squeezing has long been recognized for its numerous advantages in CV QKD, such as enhanced robustness against channel noise and loss, and improved secret key rates. However, the excess noise of the squeezed states, that unavoidably originates already from optical loss as well as other imperfections in the source, raises concerns about its potential exploitation by an eavesdropper. For the widely adopted trust assumption on the excess noise in the signal states, we confirm the stability of the protocol against the noisy squeezing in both purely attenuating as well as noisy channels in the asymptotic limit, which implies perfect parameter estimation. In the finite-size regime we show that this stability largely holds at up to 107 data points using optimal biased homodyne detection for key distribution and parameter estimation. Untrusted assumption on the noisy squeezing, on the other hand, introduces additional security bounds on the squeezing excess noise already in the asymptotic regime, which is further enforced by the finite-size effects. Additionally, we show the critical negative role of noisy squeezing in the case of atmospheric free-space channels, already in the trusted-noise and asymptotic assumptions, which emphasizes the importance of squeezing purity in the free-space quantum channels. Our results pave the way towards practical realization of squeezed-state CV QKD protocols in both fibre and free-space channels.

Quantum Science and Technology•Jan 13, 2025•0 citations

Machine learning aided carrier recovery in continuous-variable quantum key distribution

Hou-Man Chin, Nitin Jain, Darko Zibar +2 more

Abstract The secret key rate of a continuous-variable quantum key distribution (CV-QKD) system is limited by excess noise. A key issue typical to all modern CV-QKD systems implemented with a reference or pilot signal and an independent local oscillator is controlling the excess noise generated from the frequency and phase noise accrued by the transmitter and receiver. Therefore accurate phase estimation and compensation, so-called carrier recovery, is a critical subsystem of CV-QKD. Here, we explore the implementation of a machine learning framework based on Bayesian inference, namely an unscented Kalman filter (UKF), for estimation of phase noise and compare it to a standard reference method and a previously demonstrated machine learning method. Experimental results obtained over a 20-km fibre-optic link indicate that the UKF can ensure very low excess noise even at low pilot powers. The measurements exhibited low variance and high stability in excess noise over a wide range of pilot signal to noise ratios. This may enable CV-QKD systems with low hardware implementation complexity which can seamlessly work on diverse transmission lines.

arXiv:2002.09321•npj Quantum Information•Feb 4, 2021•92 citations

Back-propagation neural network based optimization for high secret key rate continuous-variable quantum key distribution systems

Yang Liu, Jiangliang Jin, Xueqin Jiang +3 more

Abstract In the continuous-variable quantum key distribution (CV-QKD) system, the secure key rate and the throughput of secure keys are important factors in measuring system quality. However, the secure key rate and the throughput of secure keys depend on various nonlinearly related parameters, which makes it difficult to effectively find the optimal parameters. In this paper, we introduce machine learning into the CV-QKD system and propose a parameter optimization scheme for the CV-QKD system based on the back-propagation (BP) neural network. The proposed scheme improves the secure key rate and throughput of secure keys by optimizing the code rate and the number of decoding iterations in the error correction process. Furthermore, under some reasonable assumptions, we prove that the optimization objective is a concave function of the optimization parameters under the BP neural network, which ensures that there exists a global optimal solution in a certain domain and provides a theoretical basis for the optimization based on BP neural networks. Simulation results show that compared with traditional optimization schemes, the proposed scheme can improve the secure key rate and transmission distance.

New Journal of Physics•Mar 27, 2026•0 citations
Abstract Parameter estimation is a critical step in continuous-variable quantum key distribution (CV-QKD), as the statistical uncertainty from a finite data size leads to pessimistic worst-case bounds that drastically reduce the secret key rate and range. While machine learning techniques have been proposed for this task, they have lacked the rigorous statistical framework necessary for integration into a composable security proof. In this work, we bridge this gap by introducing a statistically rigorous framework for using neural networks for parameter estimation in CV-QKD with quantifiable composable security. We develop a neural network estimator for the excess noise and, crucially, derive its worst-case confidence interval using a delta method approach, ensuring the estimation fails with a probability not exceeding ε PE . This allows the network to be integrated into a parameter estimation protocol that is operationally equivalent to the standard maximum likelihood method but yields significantly tighter parameter bounds. Our numerical results demonstrate that this method provides substantially more precise estimates, which directly translates into a higher secret key rate and extended transmission distance over a fiber channel under a collective Gaussian attack. This work establishes that machine learning can be securely and effectively harnessed to overcome a key performance limitation in practical CV-QKD systems.
Abstract Satellite-to-ground continuous variable quantum key distribution (CV-QKD) is constrained by rapidly varying free-space channels and short transmission windows, which limit the amount of data available for parameter estimation within each channel interval. Under finite-sample conditions, parameter point estimates are subject to statistical uncertainty and may even yield an overly favorable description of the true channel state, while errors in the estimated transmission coefficient directly affect the covariance matrix used in the security analysis. In this work, we investigate the finite-sample effects of transmission coefficient estimation in dynamic free-space CV-QKD. A neural network is introduced to estimate the transmission coefficient, and the Delta method is used to quantify the uncertainty in its predictions. The proposed method is compared with the standard maximum likelihood estimation method. Confidence intervals for the transmission coefficient are constructed for different sample sizes, and the effects of the statistical uncertainties on the secret key rate are analysed. Under the considered Gaussian-channel simulation conditions, the neural network can provide a more favorable nominal lower bound on the transmission coefficient than MLE in certain sample-limited regimes, leading to a higher parameter-estimation-limited key-rate metric. This work establishes a statistical framework that connects neural network-based transmission coefficient estimation and predictive uncertainty quantification with covariance matrix construction and secret key rate analysis. It provides a basis for machine learning-assisted parameter estimation in free-space CV-QKD systems operating with finite samples.
Continuous-variable quantum key distribution (CV-QKD) with discrete modulation has received widespread attentions because of its experimental simplicity, lower-cost implementation and ease to multiplex with classical optical communication. Recently, some inspiring numerical methods have been applied to analyse the security of discrete-modulated CV-QKD against collective attacks, which promises to obtain considerable key rate over one hundred kilometers of fiber distance. However, numerical methods require up to ten minutes to calculate a secure key rate one time using a high-performance personal computer, which means that extracting the real-time secure key rate is impossible for discrete-modulated CV-QKD system. Here, we present a neural network model to quickly predict the secure key rate of homodyne detection discrete-modulated CV-QKD with good accuracy based on experimental parameters and experimental results. With the excess noise of about 0.010.010.01, the speed of our method is improved by about seven orders of magnitude compared to that of the conventional numerical method. Our method can be extended to quickly solve complex security key rate calculation of a variety of other unstructured quantum key distribution protocols.
Abstract We address the crucial role of noisy squeezing in security and performance of continuous-variable (CV) quantum key distribution (QKD) protocols. Squeezing has long been recognized for its numerous advantages in CV QKD, such as enhanced robustness against channel noise and loss, and improved secret key rates. However, the excess noise of the squeezed states, that unavoidably originates already from optical loss as well as other imperfections in the source, raises concerns about its potential exploitation by an eavesdropper. For the widely adopted trust assumption on the excess noise in the signal states, we confirm the stability of the protocol against the noisy squeezing in both purely attenuating as well as noisy channels in the asymptotic limit, which implies perfect parameter estimation. In the finite-size regime we show that this stability largely holds at up to 107 data points using optimal biased homodyne detection for key distribution and parameter estimation. Untrusted assumption on the noisy squeezing, on the other hand, introduces additional security bounds on the squeezing excess noise already in the asymptotic regime, which is further enforced by the finite-size effects. Additionally, we show the critical negative role of noisy squeezing in the case of atmospheric free-space channels, already in the trusted-noise and asymptotic assumptions, which emphasizes the importance of squeezing purity in the free-space quantum channels. Our results pave the way towards practical realization of squeezed-state CV QKD protocols in both fibre and free-space channels.
Abstract We address the crucial role of noisy squeezing in security and performance of continuous-variable (CV) quantum key distribution (QKD) protocols. Squeezing has long been recognized for its numerous advantages in CV QKD, such as enhanced robustness against channel noise and loss, and improved secret key rates. However, the excess noise of the squeezed states, that unavoidably originates already from optical loss as well as other imperfections in the source, raises concerns about its potential exploitation by an eavesdropper. For the widely adopted trust assumption on the excess noise in the signal states, we confirm the stability of the protocol against the noisy squeezing in both purely attenuating as well as noisy channels in the asymptotic limit, which implies perfect parameter estimation. In the finite-size regime we show that this stability largely holds at up to 107 data points using optimal biased homodyne detection for key distribution and parameter estimation. Untrusted assumption on the noisy squeezing, on the other hand, introduces additional security bounds on the squeezing excess noise already in the asymptotic regime, which is further enforced by the finite-size effects. Additionally, we show the critical negative role of noisy squeezing in the case of atmospheric free-space channels, already in the trusted-noise and asymptotic assumptions, which emphasizes the importance of squeezing purity in the free-space quantum channels. Our results pave the way towards practical realization of squeezed-state CV QKD protocols in both fibre and free-space channels.
Abstract The secret key rate of a continuous-variable quantum key distribution (CV-QKD) system is limited by excess noise. A key issue typical to all modern CV-QKD systems implemented with a reference or pilot signal and an independent local oscillator is controlling the excess noise generated from the frequency and phase noise accrued by the transmitter and receiver. Therefore accurate phase estimation and compensation, so-called carrier recovery, is a critical subsystem of CV-QKD. Here, we explore the implementation of a machine learning framework based on Bayesian inference, namely an unscented Kalman filter (UKF), for estimation of phase noise and compare it to a standard reference method and a previously demonstrated machine learning method. Experimental results obtained over a 20-km fibre-optic link indicate that the UKF can ensure very low excess noise even at low pilot powers. The measurements exhibited low variance and high stability in excess noise over a wide range of pilot signal to noise ratios. This may enable CV-QKD systems with low hardware implementation complexity which can seamlessly work on diverse transmission lines.
Abstract In the continuous-variable quantum key distribution (CV-QKD) system, the secure key rate and the throughput of secure keys are important factors in measuring system quality. However, the secure key rate and the throughput of secure keys depend on various nonlinearly related parameters, which makes it difficult to effectively find the optimal parameters. In this paper, we introduce machine learning into the CV-QKD system and propose a parameter optimization scheme for the CV-QKD system based on the back-propagation (BP) neural network. The proposed scheme improves the secure key rate and throughput of secure keys by optimizing the code rate and the number of decoding iterations in the error correction process. Furthermore, under some reasonable assumptions, we prove that the optimization objective is a concave function of the optimization parameters under the BP neural network, which ensures that there exists a global optimal solution in a certain domain and provides a theoretical basis for the optimization based on BP neural networks. Simulation results show that compared with traditional optimization schemes, the proposed scheme can improve the secure key rate and transmission distance.