Federated Blockchain Learning at the Edge
Round 1
Reviewer 1 Report
This paper is motivated by the limited computational resources of the IoT devices in case of training neural networks. Then, the authors introduced a method to train the neural network at the edge to avoid restricting the data availability. The introduced method trains networks without storing any data and aggregates multiple networks, trained on unique data, forming a global network via a centralized server. After reading this paper, I have the following comments:
- Surgery and Healthcare keywords must be removed from the keywords because they are not related to the title or the contribution of the paper. Also, they mentioned few times though the paper.
- The figures quality must be improved. Some figures are in black and white, and some others are colored. Also, the size of text in the figure must be balanced.
- You must avoid adding the references as in the following statement “Therefore the most useful archetype is edge computing [9–12,16–23]”. Here 12 references are mentioned in one statement. It is not accepted.
- The organization for the rest of this paper paragraph is missing by the end of the Introduction section.
- The equations must be written using the math editor not as a normal text (Check equation 3).
- The paper is not well motivated. A motivational scenario is needed in the Introduction section. You must give strong motivation for providing your method.
Dear Prof. Editor,
The paper has some improvements to be done.
Thank you
Author Response
- Surgery and Healthcare keywords must be removed from the keywords because they are not related to the title or the contribution of the paper. Also, they mentioned few times though the paper.
- The keywords have been removed, as have some of the references to medical and surgical topics in the body of the paper.
- The figures quality must be improved. Some figures are in black and white, and some others are colored. Also, the size of text in the figure must be balanced.
- All figures are now colored cohesively such that each color has meaning.
- You must avoid adding the references as in the following statement “Therefore the most useful archetype is edge computing [9–12,16–23]”. Here 12 references are mentioned in one statement. It is not accepted.
- We now limit references to no more than two at any one statement
- The organization for the rest of this paper paragraph is missing by the end of the Introduction section.
- This section has now been added
- The equations must be written using the math editor not as a normal text (Check equation 3).
- It should be since we use latex with math mode
- The paper is not well motivated. A motivational scenario is needed in the Introduction section. You must give strong motivation for providing your method.
- The final paragraph in the introduction hopefully shows our motivation better
Reviewer 2 Report
The author proposed a distributed scheme using FL and blockchain for IoT. The idea is interesting. However, I have the following concerns.
1. Please revise the grammatical issues of the paper.
2. Recent Blockchain-based FL papers are missing. Some are mentioned as follows.
-> "Trustworthy Federated Learning via Blockchain," in IEEE Internet of Things Journal, vol. 10, no. 1, pp. 92-109, 1 Jan.1, 2023, doi: 10.1109/JIOT.2022.3201117.
-> "FBI: A Federated Learning-Based Blockchain-Embedded Data Accumulation Scheme Using Drones for Internet of Things," in IEEE Wireless Communications Letters, vol. 11, no. 5, pp. 972-976, May 2022, doi: 10.1109/LWC.2022.3151873.
3. Blockchain-based FL is a very popular topic. Please highlight your novelty based on the limitation of existing works.
4. It would be better to include mist and dew computing in the discussion.
5. Tensorflow Lite is just a tool. Why did the authors highlight it in the technical details?
6. Motivation is not clear. Moreover, the motivation for using blockchain is missing in the introduction.
7. Technical depth is missing. Without that paper seems like an experimental report.
8. An analysis of computational complexity and convergence is required.
9. An analysis with existing works is missing and a comparison is required in the performance analysis section.
Needs to improve.
Author Response
- Please revise the grammatical issues of the paper.
- We have attempted to do so
2. Recent Blockchain-based FL papers are missing. Some are mentioned as follows.
-> "Trustworthy Federated Learning via Blockchain," in IEEE Internet of Things Journal, vol. 10, no. 1, pp. 92-109, 1 Jan.1, 2023, doi: 10.1109/JIOT.2022.3201117.
-> "FBI: A Federated Learning-Based Blockchain-Embedded Data Accumulation Scheme Using Drones for Internet of Things," in IEEE Wireless Communications Letters, vol. 11, no. 5, pp. 972-976, May 2022, doi: 10.1109/LWC.2022.3151873.
Two excellent paper, thank you for bringing them to our attention
3. Blockchain-based FL is a very popular topic. Please highlight your novelty based on the limitation of existing works.
This should be shown in the two subsections Related Works and Contributions
4. It would be better to include mist and dew computing in the discussion.
Indeed, these have been added too, especially mist
5. Tensorflow Lite is just a tool. Why did the authors highlight it in the technical details?
It was mentioned due to its ability to allow users to automatically reap the benefits if they have a tensorflow model with little to no overhead.
6. Motivation is not clear. Moreover, the motivation for using blockchain is missing in the introduction.
Hopefully this has been made clear by the last paragraph in the Introduction
7. Technical depth is missing. Without that paper seems like an experimental report.
We have had trouble with this given the time frame
8. An analysis of computational complexity and convergence is required.
We have also had trouble with this given the time frame
9. An analysis with existing works is missing and a comparison is required in the performance analysis section.
This has been added in the section Related Works.
Round 2
Reviewer 1 Report
Good job. only the figures need to seem like scientific figures. You can use MS Visio, Draw-io, or any other drawing software to produce better figures.
Dear Prof. Editor,
The authors handled most of my comments. But the figures must be updated to seem as scientific figures.
Thank you
Author Response
Good job. only the figures need to seem like scientific figures. You can use MS Visio, Draw-io, or any other drawing software to produce better figures.
- I've tried to pick less garish colours but I'm not too sure how to improve the figures. I've tried to style them like Fig 1 in this amazing paper: Federated Learning With Blockchain for
Autonomous Vehicles: Analysis
and Design Challenges. Would you be able to point me in the direction of the kinds of figures you recommend?
Minor editing of English language required
- I am unsure what editing is needed, if possible could you show me an example? Is my writing not formal/scientific enough?
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Thank you very much for considering my comment. However, there are some comments that is handled but changes are not satfactory.
7. Technical depth is missing. Without that paper seems like an experimental report.
Authors response -> We have had trouble with this given the time frame
Please hightlight changes regarding this.
8. An analysis of computational complexity and convergence is required.
Authors response -> We have also had trouble with this given the time frame
-> I didn't find any section regarding this.
Minor changes are required.
Author Response
7. Technical depth is missing. Without that paper seems like an experimental report. Please hightlight changes regarding this.
- We have added the below change as a new section in order to add technical depth. However, I am unsure that this is enough to fully satisfy the missing level of detail. If it is possible, could you give me an example of where detail is lacking and I would be very happy to improve the paper in any way possible.
8. An analysis of computational complexity and convergence is required.
-> I didn't find any section regarding this.
- New section, #3 System complexity has been added.
Author Response File: Author Response.pdf
