Backpack Client Selection Keeping Swarm Learning in Industrial Digital Twins for Wireless Mapping

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper proposes a backpack model-based approach for wireless mapping modeling of digital twins in industrial IoT, and examines its convergence, model error, and parameter transmission. The paper has strong innovation, but there are still some issues：

1. The paper assumes that the loss function satisfies continuous differentiability and has strong convexity, but does not explain the specific reasons and rationality for proposing this assumption.
2. In the construction of the backpack model, there is a lack of differentiation from traditional FL client selection methods. Please explain.
3. In Equation 11, the model parameters are weighted average updated, but the left end of the equation uses the model parameter θ(t+1) i of device i. Please provide a detailed explanation.
4. Equation 14 proposes the inequality that gradient descent error should satisfy, equation 16 proposes the inequality that the global model error of SL network should satisfy, the similarities and differences between the two equations should be explained in detail.

Author Response

Thank you very much for your suggestions, which are crucial for us to improve the quality of our manuscript.Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

- A more detailed explanation of 3. system modal, including the formulation and parameters, is necessary. It would also be helpful to include a table clearly summarizing the meaning of each component in the formulation.
- In the conclusion, the paper should additionally address the contributions of the study, a summary of the results, limitations, and suggestions for future research.

Author Response

Thank you very much for your suggestions, which are crucial for us to improve the quality of our manuscript.Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors,

the article provides useful contributions in swarm learning for digital twins in industrial IoT. A new DT-SL architecture is proposed. The basic algorithm for client selection is based on the knapsack optimization model. Significant improvement in convergence parameters, MSE and amount of transmitted data is demonstrated.

Negative:

The conclusion is concise. I suggest expanding the conclusion with a broader summary of the results achieved by the authors, with an emphasis on the contributions and potential for further development, or their further research in this area. The article lacks a Discussion (or could also be part of the Conclusion), which would clarify the contributions of the authors from their site in comparison with current practice. The weaker side is a less detailed description of the implementation process, which could be part of a comment within the Discussion section.

I hope that the comments will increase the quality of your contribution.

Best regards

Author Response

Thank you very much for your suggestions, which are crucial for us to improve the quality of our manuscript.Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This paper presents the DT-KSL method and compares the findings with other methods in the literature. 

There are several issues within the manuscript (e.g. spacing, mathmode, etc.) that need to be corrected. Additionally some variables like DP() could be represented by a single variable to improve readability. 

Lastly, it is not clear how much better this method is over other methods. Additionally, the backpack model is not described. Also, the convergence proof and mathematical representations lack rigor. It is not clear about some of the sets, or what dimensions are, etc. 

Please see PDF marks for your review. 

Comments for author File: Comments.pdf

Author Response

Thank you very much for your suggestions, which are crucial for us to improve the quality of our manuscript.Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Dear Authors,

I think that you made a corrections that improvrd the quality of your article.

I hate no another comments.

Best regards 

 

 

Reviewer 4 Report

Comments and Suggestions for Authors

no further comments.