Adjusting Optical Polarization with Machine Learning for Enhancing Practical Security of Continuous-Variable Quantum Key Distribution

Round 1

Reviewer 1 Report

·        General: It is recommended that a schematic diagram be added, illustrating the approach for optimal dynamic polarization control. Likewise, it is recommended to add a graphical abstract.  

·        Novelty: It is unclear why this approach is novel to achieve the optimal dynamic polarization control. For instance, machine learning was suggested, what is the base line for the optimized solution such that machine learning can show advantage or supremacy.   

·        Define Key Terms: There is a need to precisely define “Optimal,” and how the machine learning address current challenges.  

·        Quantify Improvement: Identify and quantify the improvement achieved in terms of approach (i.e., computation cost, including training, size of data, etc.) and system (i.e., higher fidelity, lower error rate, etc.). Please be advised that Fig. 4 alone is not usually sufficient.   

·        Reproducibility and Repeatability: you need to provide full information about the machine learning models including but not limited to data set (i.e., training, testing, and verification), package used, programming language, etc. Also, you need to share the code/data.   

·        Limitations: Acknowledge any limitations of your approach and suggest potential future directions for research.

·        Equations: please review and cite when appropriate. For instance, eq 22, is this an equation derived by you, if yes, please confirm in the manuscript by saying “we define over-noise caused by the optical leakage of the actual principal vibrations of the DPC as…,” otherwise, you need to add a reference.

·        Figures and Tables: Please review all, for instance, Fig.1 does not include a figure caption to subfigure (b).

English language is fairly adequate. 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 2 Report

In the paper, the authors introduce an optimized dynamic polarization control (DPC) scheme leveraging machine learning algorithms to enhance the practicality and security of Continuous Variable Quantum Key Distribution (CVQKD) systems.

The central question the authors tackle is how to optimize the dynamic polarization control of quantum signals in CVQKD systems to counteract the adverse effects of atmospheric turbulence, thereby preventing light leakage and improving the system's secret key rate. By employing machine learning to dynamically adjust the polarization control in response to the fluctuating polarization states of transmitted quantum signals, the authors propose a solution to mitigate light leakage. The importance of this work lies in its potential to overcome a critical barrier to the practical implementation of secure quantum communications, paving the way for more robust and widely applicable quantum cryptographic systems.

I find the paper interesting and well-presented. However, I find the lack of machine learning method description and comparison. Also, there are pioneering works on machine learning for design of experiments, espicially in optics, optical networking, and QKD:

https://doi.org/10.1073/pnas.1714936115
https://link.aps.org/doi/10.1103/PhysRevLett.125.160401
https://link.aps.org/doi/10.1103/PRXQuantum.1.010301

Please discuss these works is details in the paper.

Author Response

Please see the attachment

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript proposes an enhanced machine learing (ML)-based dynamic polarization control (DPC) for optimizing practical continuous quantum distribution (CVQKD) systems over fading channels with randomly fluctuating tranmittance. The ML-based DPC can be improved in terms to better handle polarization deteriorations of transmitted quantum signals in fading channels by employing an ML-involved optimization algorithm that delivers precise responses to abnormal transformations of polarization states. Furthermore, numerical simulations confirms the robustness and adaptability of the ML-based DPC. The authors states that the proposed scheme effectively suppresses light leakage, preserving the integrity of deteriorated quantum signals and enhancing the reliability of practical CVQKD systems.

The findings have significant implications for the advancement of Machine Learning Assisted Continuous-Variable Quantum Key Distribution.  

As the minor suggestions, authors may also address in the introduction to a wider audience, by saying that non-markovianity is useful to enhance security, see eg Vasile  et al "Continuous variable quantum key distribution in non-Markovian channels", Phys. Rev. A 83, 042321 (2011)
and then saying that pbg and Lorentzian media are promising to obtain nonMarkovian behaviour, i.e. the demands for experimental study and its implication of your results for recent development in quantum technology and on photonic bandgap that may provide a reliable way to transmit entanglement over long distance and potential application with security (Sci Rep 12, 11646 (2022). https://doi.org/10.1038/s41598-022-15865-5) and on Lorentzian environment like in (https://doi.org/10.1016/j.physleta.2022.128022) and how phase modulation of coherent states plays in quantum technology https://doi.org/10.1088/0031-8949/2010/T140/014062 and the use of probabilistic noiseless linear amplifiers both at the encoding stage https://doi.org/10.1364/JOSAB.36.002938 where the information is coded on phase shifts and at the decoding stage https://journals.aps.org/pra/abstract/10.1103/PhysRevA.93.062315  the authors are expected to raise the level of the manuscript by referring to them appropriately.

I think this paper is a worthy contribution for Applied Sciences, and I recommend it for publication after the above remarks are addressed. 

Author Response

Please see the attachment

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The novelty of the proposed Machine Learning approach is unclear. It needs to be elaborated further.

English language level is adequetly suffcient. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

My questions were addressed

Author Response

Please see the attachment.

Author Response File: Author Response.docx