Next Article in Journal
Stochastic and Extreme Scenario Generation of Wind Power and Supply–Demand Balance Analysis Considering Wind Power–Temperature Correlation
Next Article in Special Issue
DiffPlate: A Diffusion Model for Super-Resolution of License Plate Images
Previous Article in Journal
Interspectral Error Tracking and Compensation of DSDT in Satellite Internet of Things
Previous Article in Special Issue
EEG_DMNet: A Deep Multi-Scale Convolutional Neural Network for Electroencephalography-Based Driver Drowsiness Detection
 
 
Article
Peer-Review Record

An Integrated DQN and RF Packet Routing Framework for the V2X Network

Electronics 2024, 13(11), 2099; https://doi.org/10.3390/electronics13112099
by Chin-En Yen 1, Yu-Siang Jhang 2, Yu-Hsuan Hsieh 2, Yu-Cheng Chen 2, Chunghui Kuo 3 and Ing-Chau Chang 2,*
Reviewer 1:
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Electronics 2024, 13(11), 2099; https://doi.org/10.3390/electronics13112099
Submission received: 22 April 2024 / Revised: 19 May 2024 / Accepted: 26 May 2024 / Published: 28 May 2024
(This article belongs to the Special Issue Signal Processing and AI Applications for Vehicles)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper highlghts the significance of deep reinforcement learning (DRL) in designing intelligent vehicle-to-everything (V2X) routing protocols for vehicular ad-hoc networks (VANET) with the development of artificial intelligence technology. The paper provides a comparative analysis with existing routing algorithms (TDRLRP, VRDRT, and traditional VANET routing algorithms) demonstrates significant performance improvements in terms of average packet delivery ratio, end-to-end delay, and overhead ratio for both sparse and congested periods through intensive simulations.

 

However, following comments shall be addressed:

  • The manuscript should accurately describe the authors' contribution. For example, instead of stating "we have proposed the Software-Defined Directional QGRID (SD-QGRID) routing architecture [11]," it should be clarified that the authors have adapted, not proposed, this architecture. Careful revision is necessary to highlight the novelty of the paper and precisely specify what has been proposed.

  • Regarding the statement "The DQN network architecture diagram of the SDN-enabled vehicular network," it's important to ascertain whether similar approaches utilizing DQN have been employed in related works. Additionally, the authors are encouraged to provide a comprehensive review highlighting key differences between their proposed approach and existing literature.

  • In the statement "In the offline phase, we utilize real-world moving trajectory data of Beijing taxis and propose algorithms to correct vehicle GPS trajectories," it would be beneficial to elaborate on these algorithms. Are they novel algorithms? Why were they chosen for use? Providing further details would enhance the understanding of the methodology employed.

  • While the paper's contribution is well-introduced, the related work section could be improved by delineating key differences with similar works in the field.

  • Typos, such as missing "a" or "an," should be corrected throughout the manuscript.

 

 

Comments on the Quality of English Language

Minor editing of English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The authors in the manuscript entitled “An Integrated DQN and RF Packet Routing Framework for the V2X Network” combined DQN and RF approaches and added a step to process the vehicle historical trajectory data to train the model. They have compared their results with other protocols and presented results. The work is promising but the authors need to consider the following.  

1.     Abstract is very confusing and authors need to rewrite.  

2.     Add numerical results in abstract.

3.     Use the full text and abbreviation when the text is used first time and then use abbreviation only.

4.     Add more references in related work section.

5.     It is recommended to separate all formulae from text and assign numbers. See section 3.1.2.

6.     Use better quality pictures and use light colors.

7.     The axes should be same for figures 30, 31. Same for figures 32, 33 and 34, 35 and rest of the graph pairs.

8.   In conclusion section, the statement “In the simulations, IDRF consistently outperforms other algorithms” must be supported with numerical results from previous sections.

9.     Some references are old. Replace with recent references.

Comments on the Quality of English Language

Quality of English must be improved. Some sections are confusing and lack proper flow.   

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

 The paper is interesting but needs revision. The authors are asked to consider the following points

+ The “Introduction” section has too much detail on DRL & DQN. These contents can be organized in a new section

+ The contribution bullet points in the “Introduction” are too long and they do not explicitly say what new research is in them.

+ The quality of the figures is not good, such as Fig. 1

+ Section 2 (Related Works) is poorly written. The authors simply start each paper in a bullet point and discuss those papers. This is not a proper literature review. The works need to be organized in different concepts as they progress in time and compare and contrast those works, citing their strength and limitations. This section needs to be revisited extensively. Also, the following papers are relevant to this work in networking, and ML aspects which the authors might consider: https://doi.org/10.3390/s24041216; https://doi.org/10.1109/TITS.2021.3059261; https://doi.org/10.1109/TITS.2021.3053958; https://doi.org/10.1109/CSDE53843.2021.9718396

+ Some basic information is redundant, like the difference between the decision trees and RF which can be found in any book on ML. Similarly, the pseudocode for RF in Fig. 11.

+ From the vehicle trajectory data, the authors create another dataset by including vehicle stopping time at the intersection. Obviously, there will be some margin of error in this calculation, as the stopping time at the intersection depends on congestion, which direction you are moving and at what speed you are moving. This error will be gradually accumulated in other calculations as well. This review is not convinced whether the calculation is correct through various steps.

+ The paper has too many diagrams, some basic figs are not and some (such as Fig. 26) can be simplified.

+ What mobility model has been used in the simulation? Why the authors did not use a traffic simulator like SUMO (https://eclipse.dev/sumo/)?

 

+ From the writing style of the submission, it seems that a thesis chapter can be converted into a paper, but it has not got the proper shape of the paper.

Comments on the Quality of English Language

Please check the paper properly for the quality of the English language.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have implemented the recommended changes

Comments on the Quality of English Language

Minor editing is required.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have addressed all my concerns in the revised manuscript. In my opinion, the manuscript is now suitable for publication.

Back to TopTop