All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
We have now received the input from the experts on your revised article. Based on their feedback, I am happy to let you know that your manuscript has been recommended for publication. Congratulations!
[# PeerJ Staff Note - this decision was reviewed and approved by Yilun Shang, a 'PeerJ Computer Science' Section Editor covering this Section #]
NA
NA
Authors addressed the reviewers comments.
Authors addressed the reviewers comments.
All my comments have been incorporated
All my comments have been incorporated
All my comments have been incorporated
All my comments have been incorporated
We have carefully reviewed your paper and you will see that the reviewers and recommending a lot of improvements, so we would suggest you update your paper carefully and re-submit.
Editor Comments: Please Consider revising for readability, especially for readers unfamiliar with the technical jargon.
The paper introduces several key concepts, such as U-net, fuzzy reinforcement Q-learning, and fuzzy neural networks, but doesn't explain them in sufficient detail.
the paper could include a more in-depth comparative analysis with other existing channel equalization methods.
It would be valuable to include potential real-world scenarios where the U-FRQL-EA algorithm could be applied
Technical language of the paper should be improved.
Please mention the computational overhead with fuzzy reinforcement learning. It is important as the UAV has limited power.
The robustness of the algorithm should be checked through evaluating with varying noise conditions.
The proposed U-FRQL-EA was compared with CNN-FA and RBF algorithms, However, there should be a comparison with more recent techniques, such as techniques that are based on deep learning or hybrid approaches.
There is no justification for why U-FRQL-EA converged faster than RBF or CNN-FA.
The results are simulation-based. There should be Real-world experimentation, or the authors should use real-world data to enhance the validity of the results.
The performance of BER is evaluated in normal conditions. However, there is a need to check the robustness of the algorithm in extreme interference or noise conditions.
The computational complexity to train the U-net or the fuzzy Q-learning components should be mentioned because the system is expected to work in real time environment.
The nodes reduce their fuzzy value when packet forwarding fails, however, no detail has been provided on handling network congestion.
The specific constraints are not addressed in paper such as limited battery life. How does the proposed algorithm consider these limitations?
The security issues are not addressed in manuscript such as signal jamming.
There should be a sensitivity analysis to show how the number of nodes or noise levels affects the performance of the algorithm.
The simulation system is not generalized as in real-time the distance can be more than 200 meters.
The performance of BER is evaluated in normal conditions. However, there is a need to check the robustness of the algorithm in extreme interference or noise conditions.
The computational complexity to train the U-net or the fuzzy Q-learning components should be mentioned because the system is expected to work in real time environment.
The simulation system is not generalized as in real-time the distance can be more than 200 meters.
Please mention the computational overhead with fuzzy reinforcement learning. It is important as the UAV has limited power.
The robustness of the algorithm should be checked through evaluating with varying noise conditions.
The proposed U-FRQL-EA was compared with CNN-FA and RBF algorithms, However, there should be a comparison with more recent techniques, such as techniques that are based on deep learning or hybrid approaches.
There is no justification for why U-FRQL-EA converged faster than RBF or CNN-FA.
The results are simulation-based. There should be Real-world experimentation, or the authors should use real-world data to enhance the validity of the results.
The performance of BER is evaluated in normal conditions. However, there is a need to check the robustness of the algorithm in extreme interference or noise conditions.
The computational complexity to train the U-net or the fuzzy Q-learning components should be mentioned because the system is expected to work in real time environment.
The nodes reduce their fuzzy value when packet forwarding fails, however, no detail has been provided on handling network congestion.
The specific constraints are not addressed in paper such as limited battery life. How does the proposed algorithm consider these limitations?
The security issues are not addressed in manuscript such as signal jamming.
There should be a sensitivity analysis to show how the number of nodes or noise levels affects the performance of the algorithm.
The simulation system is not generalized as in real-time the distance can be more than 200 meters.
The paper presents a new equalization algorithm that combines a U-net model with fuzzy reinforcement Q-learning to boost the channel sensing and equalization abilities of UAV communication systems.
To make the paper clearer and more impactful, here are a few suggestions to consider:
1- There exist typos in manuscript i.e. in line no. 33, there should be "uncrewed" in place of "unscrewed". Similarly, in funding statement, the word "no" should be replaced with "not". Authors are suggested to revise the manuscript.
2- There is inconsistency in terminologies such as "U-net" and "U-Net".
3- in section 3.1, there is no justification for selecting the kernel sizes(1*4,1*5). Please provide the proper reasoning and should explain that how these sizes improved the performance.
4- 16 layers are used in U-net model, however, the authors are required to provide the its justification.
5- Authors are suggested to provide the noise characteristics of UAV communication channels.
6- In fuzzification, the parameters are not clearly defined such as membership functions. Please provide the criteria of selecting the parameters for fuzzification.
7- The hyperparameter tuning for the U-FRQL-EA algorithm has not been provided, such as learning rate, network architecture etc.
8- Please describe the scalability of U-FRQL-EA algorithm as it is tested on a limited number of nodes. Will the algorithm accommodate the larger number of nodes.
9- It has not been mentioned how much fuzzy logic contributed to performance improvement. There should be a comparison with the non-fuzzy Q-learning technique.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.