Sea-Based UAV Network Resource Allocation Method Based on an Attention Mechanism
, Faping Lu 1, Yaozong Pan 1, Tianqi Zhang 2, Jiafang Kang 1, Zhiyong Zhao 1 and Yang You 3
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe reviewer has the following concerns.
1) The reviewer has no idea about the first few sentences in the Introduction.
2) It is inappropriate to write phrases in past tense in the part Contribution.
3) Lots of jargon/abbreviations are involved, which should have been listed in a table for ease of following.
4) In the sentence "The current research on wireless network resource allocation focuses on three aspects: power allocation[5], spectrum allocation[6], and access selection.", spaces are missing before the citations. This issue exists in other parts of this manuscript.
5) The literature review regarding recent advances in the field of DRL-aided resource allocation and performance optimization for UAV-aided networks is not as comprehensive as anticipated. More phrases are expected to be used to deliver a more proper overview of DRL-aided UAV-mounted networks. For example, “Path planning for cellular-connected UAV: A DRL solution with quantum-inspired experience replay” and “Radio resource management for cellular-connected UAV: A learning approach”.
6) Figures are advised to be replaced with their HD versions, e.g., in eps format.
7) An explicit complexity analysis of the proposed approach is needed, to make the technical contents more complete.
8) Convergence performance of the proposed algorithm is expected to be further discussed.
9) Why are there leaps around Episode 20 in Fig. 7, for the top 3 curves?
10) Any strategies used to tackle potential sparse/delayed reward issues?
11) Could the proposed algorithm be extended to multi-agent systems?
12) The fastest rotary-wing UAV maximumly files with 225 km/h, i.e., 62.5 m/s. How come the max high-speed node is set to 150 m/s?
13) Does the Doppler frequency shift effect be considered? If not, then why?
14) What are the intuitive reasons that the proposed algorithm can outperform baselines?
15) What is the value of noise in simulation and what is that of the noise power spectral density?
Comments on the Quality of English Languagemore efforts in proofreading are needed, as there are lots of minor issues in English writing and phrasing
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for Authors
In this article, the authors aimed to enhance the alignment between network resources and node services by introducing an attention mechanism and Double DQN (DDQN) algorithm that optimizes the service access strategy, curbs action outputs, and improves service-node compatibility. The algorithm constitutes a method for UAV network resource allocation in marine environments. The author’s work is not bad, but there are still several issues that must be considered, such as:
1. First of all, the contributions of this study added should be refined, avoid unnecessary discussion and keep the main discussion and focus of this paper.
2. The methodology part is intricate and lacks adequate lucidity, particularly in the elucidation of the deep reinforcement learning methods employed. The authors should enhance the level of granularity in their descriptions of the algorithmic process and provide explicit details regarding the exact parameters employed.
3. There should be a more rigorous justification for choosing the Double DQN (DDQN) and other deep reinforcement learning approaches. The authors should do a comparative analysis of these approaches with other viable algorithms and present a well-founded justification for why DDQN is the most appropriate choice for this application, specifically in maritime environments.
4. The simulation setup and parameters are given briefly but lack thoroughness. The authors should elaborate on the methodology used to choose the simulation parameters and include further contextual information regarding their significance in relation to practical maritime UAV networks. Also, it is essential to incorporate sensitivity analysis on crucial parameters to showcase the flexibility of the results.
5. The literature review is weak. Therefore, the authors should consider some more up-to-date literature for better understanding. The authors can also refer to “A UAV-Swarm-Communication Model Using a Machine-Learning Approach for Search-and-Rescue Applications”, “Mobile Robot Localization: Current Challenges and Future Prospective”, and so on.
6. Although the paper briefly compares its results with previous methodologies, this comparison lacks depth and thoroughness. The authors should do a comprehensive analysis, comparing the proposed method with both traditional and machine learning-based approaches that are currently considered the best in the field.
7. The material is based exclusively on simulated data. However, to enhance the credibility of the results, it is recommended that the authors provide a discourse regarding the performance of the suggested approach when applied to real-world data. If there is a lack of real-world data, it is important to recognize this restriction publicly and propose prospective future studies.
8. The paper fails to address the constraints of the proposed approach sufficiently. The authors ought to present a more comprehensive examination of possible vulnerabilities, such as the dependence on certain environmental circumstances or assumptions made during modelling and the potential impact of these on the applicability of the results.
9. The textual analysis of the simulation results, as depicted in Figures 7 and 8, lacks appropriate interpretation. The authors should enhance the level of depth in their discussion regarding the implications of these figures on the performance of the suggested approach, particularly when compared to the other algorithms that were examined.
10. There are multiple grammatical mistakes and typo errors. The manuscript would benefit from a comprehensive revision to enhance its readability and ensure that all technical words are precisely defined and consistently employed throughout the text.
11. The existing references are too limited and not enough. The authors should cite some more latest articles for better comparison.
Comments on the Quality of English LanguageExtensive editing of English language required.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors have responded to my comments.
Reviewer 2 Report
Comments and Suggestions for AuthorsNA
