DDPG-Based Computation Offloading Strategy for Maritime UAV

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The topic is important and, in time, edge computing and UAV-assisted MIoT are emerging as hot topics, particularly in remote sensing, autonomous maritime operations, and low-latency networks. The integration of DDPG with joint optimization (offloading ratio, trajectory, and resource allocation) is methodologically interesting and aligns with current trends in RL-based networking. Although limited, the simulation demonstrates the algorithm’s potential under idealized conditions, providing a good first step.
However, several important concerns must be addressed before this work can be considered for publication.
Major concerns:
- The manuscript lacks real-world data, field validation, or realistic modeling of maritime conditions. Marine environments pose complex wireless and physical dynamics that are abstracted away, limiting the relevance and credibility of simulation-based results;
- The proposed DDPG-based approach is not compared with strong contemporary algorithms such as MADDPG, PPO, or even heuristic or convex optimization approaches. A meaningful performance assessment requires comparison to state-of-the-art alternatives;
- Key modeling assumptions (e.g., wireless link quality over the sea, energy consumption models, task arrival patterns) are vaguely described. The reward function, constraints, and trajectory planning mechanism are insufficiently explained for reproduction;
- The contribution of each component (offloading decision, trajectory, resource allocation) is not isolated or analyzed. Sensitivity to parameters such as task size or UAV energy should be evaluated;
- The conclusion claims robustness in “highly dynamic maritime environments,” but no stochastic variability or adversarial testing is provided in the simulations to support this;
- The manuscript does not address the limitations of the approach (e.g., UAV endurance, communication latency, risk of packet loss, DDPG convergence stability), nor does it discuss scalability to multi-UAV or multi-edge-node scenarios. As we can see from the latest maritime battles (e.g., during the Russian-Ukrainian conflict), the fleet of drones - is the very promising thing.

Overall recommendations that I could give are as follows:
- Improve the technical depth of modeling and clearly formalize the optimization problem;
- Include a thorough benchmarking with modern baseline approaches;
- Expand the literature review to cover state-of-the-art works;
- Perform additional evaluations to test generalizability and robustness;
- Revise language and improve the structure and clarity of algorithm descriptions.

Comments on the Quality of English Language

Language needs polishing (e.g., grammar, clarity, sentence construction).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This article proposes a novel computing offloading and resource allocation algorithm for maritime UAVs based on DDPG. It jointly optimizes task offloading ratio, UAV trajectory planning, and edge computing resource allocation under energy constraints, aiming to minimize the overall system delay while satisfying energy limitations. The effectiveness and robustness of the proposed DDPG strategy are demonstrated through simulations in dynamic maritime environments.

1. Many words are unnecessarily separated by “-”. Please check throughout the paper.
2. Line 445, the sentence is not finished after “when”
3. The title focuses on maritime environment, while the content of the paper does not specify the characteristics of the maritime environment, such as high dynamic changes, limited energy, high transmission delay, delay sensitivity, etc. , which are not discussed and appear in the experiment setup.
4. Especially, the dynamic environment change could be the main difference from traditional city environment and should be discussed and given more related parameter setup.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors propose a DDPG-based computation offloading strategy in the context of maritime internet of things. The proposed method is validated by means of simulations and its performance its compared with other existing approaches. The authors should take into account the following comments and suggestions in order to improve the overall quality of the paper.

Comments regarding technical aspects:

• The main contributions of the paper should be clearly highlighted towards the end of Section 1; A remainder detailing the contents for the rest of the paper should be also inserted at the end of Section 1;
• Some comments regarding such existing hardware (UAVs with MEC server capabilities) would be necessary in order to illustrate the feasibility of the proposed approach;
• Details regarding the way in which the synchronization should be made, as a time division approach is used in the context of the proposed system;
• The angle θ (row 139) is not clearly defined; the H parameter in eq. 2 is also not defined; Fig. 1 should be improved by adding the above-mentioned parameters, for a clearer understanding of the scenario;  
• The total latency should be more clearly defined (rows 164-165);

Comments regarding editing aspects:

• Some abbreviations are not explained the first time when they are used in the text (M-IoT on row 36, UAV on row 50, QoS on row 51, DDPG on row 53, SDN on row 73, NP row 242, aso); even if the abbreviations are explained in the abstract, they should be also explained the first time when they are used in the text;
• A unitary abbreviation should be used all over the paper, either M-IoT or MioT;
• Sections shouldn’t start directly with a subsection (see Section 3);

Comments regarding grammar/typos:

• …via UAV edge nodes… instead of current form (rows 20-21);
• …satisfying… instead of …satisfy… (row 23);
• Computational model instead of Calculation model (row 158).

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear authors, thanks for your work.

I'd recommend this paper for publication after the authors take care of this comment as well:

• Expand the literature review to cover state-of-the-art works;

(The statements on lines 52-88 are to be supported by the references, and the Introduction can be enhanced in general)

Author Response

Reviewer 1:

Thank you for your comments and constructive suggestions. Your valuable time paid for this paper is very appreciated. According to the comments, this manuscript has been carefully revised and improved.

Comment 1: The statements on lines 52-88 are to be supported by the references, and the Introduction can be enhanced in general.

Response 1:

Prevalent salt spray and fog introduce variable attenuation and absorption of radio waves [7].

7. Yan, C.; Fu, L.; Zhang, J.; Wang, J. A Comprehensive Survey on UAV Communication Channel Modeling. IEEE Access 2019, 7, 107769–107792, doi:10.1109/ACCESS.2019.2933173.

Relevant content added to lines 57-58.

..., including rapid deployment, flexible mobility, and line-of-sight communications [8].

8. Akhtar, M.W.; Saeed, N. UAVs-Enabled Maritime Communications: UAVs-Enabled Maritime Communications: Opportunities and Challenges. IEEE Syst. Man Cybern. Mag. 2023, 9, 2–8, doi:10.1109/MSMC.2022.3231415.

Relevant content added to lines 71-72.

UAV-assisted maritime MEC networks are capable of meeting the computational demands of user devices, while concurrently reducing energy consumption and task latency [9].

9. Jiao, X.; Chen, Y.; Chen, Y.; Wu, X.; Guo, S.; Zhu, W.; Lou, W. SIC-Enabled Intelligent Online Task Concurrent Offloading for Wireless Powered MEC. IEEE Internet Things J. 2024, 11, 22684–22696, doi:10.1109/JIOT.2024.3383469.

Relevant content added to lines 81-82.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have properly addressed all my previous comments and suggestions.

A minor editing issue is the fact that the last three references are not complete, all details regarding them should be added.

Author Response

Reviewer 3:

Thank you for your comments and constructive suggestions. Your valuable time paid for this paper is very appreciated. According to the comments, this manuscript has been carefully revised and improved.

Comment 1: A minor editing issue is the fact that the last three references are not complete, all details regarding them should be added.

Response 1:

Thank you for your comment. Based on your feedback, the literature content has been revised and completed.

21. Lee Y H , Dong F , Meng Y S .Near Sea-Surface Mobile Radiowave Propagation at 5 GHz: Measurements and Modeling[J].Radioengineering, 2014, 23(3):824-830.DOI:10.1002/etep.1788.
22. Hasselt H V , Guez A , Silver D .Deep Reinforcement Learning with Double Q-learning[J].Computer ence, 2015.DOI:10.48550/arXiv.1509.06461.
23. Wang Z , Freitas N D , Lanctot M .Dueling Network Architectures for Deep Reinforcement Learning[J].JMLR.org, 2015.DOI:10.48550/arXiv.1511.06581.