Deep Reinforcement Learning-Based Intelligent Security Forwarding Strategy for VANET
Round 1
Reviewer 1 Report
paper titled "Deep Reinforcement Learning Based Intelligent Security Forwarding Strategy for VANET" addressed an important research problem in practice. The authors presented the well defined introduction about the considered problem. The related work is sufficient and defined the existing
methodologies compared and defined well. There are many weaknesses in the work.
1. The time complexity of the proposed algorithms need to be addressed in detail
2. There are many notations in the manuscript, however there must be table of the notation in the manuscript.
3. The proposed method for VANET seems not to be adaptive, what happends if there are changes in the environment during processing in the network
4. The simulation performance comparison is missing with any existing methods. Need detailed performance comparison with any one of the methods discussed in table 1/2
5. authod should add detailed computational complexity analysis with existing methods.
6. Paper needs proof reading. It contains many grammatical mistakes.
7. reference need improvement as research developement and work of year 2022 is missing in references
Author Response
The authors would like to thank the Editor and the Reviewers for their precious time and constructive comments and suggestions. Those comments are all valuable and very helpful for revising our paper, as well as the important guiding significance to our researches. We have carefully addressed all the comments. Following this letter are the Reviewers’ comments in black normal font with our responses in red font. Meanwhile, changes made to the original manuscript are highlighted in the revised one. The revision has been developed in consultation with all coauthors, and each author has given approval to the final form of this revision. Our responses are given in a point-by-point manner below.
Point 1: The time complexity of the proposed algorithms need to be addressed in detail.
Response 1: Thanks for the Reviewer’s kind suggestion. We completely agree with the comment made by the Reviewer and really appreciate for pointing this out. We add the complexity analysis of the proposed method. At the same time, the complexity of the comparison method (Q-LBR) is also analyzed, so as to illustrate the advantages of the proposed method. The changes are made on pages 8-9, lines 292-305.
Point 2: There are many notations in the manuscript, however there must be table of the notation in the manuscript.
Response 2: Thanks for the Reviewer’s kind suggestion. We have added the table of notations(see Table 1 for details).
Point 3: The proposed method for VANET seems not to be adaptive, what happends if there are changes in the environment during processing in the network.
Response 3: Thanks for your question. We think our method is adaptive. In this paper, for the convenience of explanation, message types are summarized into the combination of urgent and reliable dimensions, which are represented by and . We set and to 1 or 30 during simulation. In practical applications, we could flexibly set and based on the message types. We could also expand the value range to better match environmental requirements. After parameter setting and model training, the method could be self-adaptive. In addition, the state space of the proposed method also includes the distance, delay and security attributes of nodes.
Point 4: The simulation performance comparison is missing with any existing methods. Need detailed performance comparison with any one of the methods discussed in table 1/2.
Response 4: Thanks for the Reviewer’s kind suggestion. According to your suggestion, we use Q-LBR as a comparative method. This is because Q-LBR is the most similar to our proposed method in all respects. We analyzed the performance of the two methods with three message types. The changes are made in lines 324-332, equation 7, lines 409-416, figure 4, lines 459-467, and table 4.
Point 5: Author should add detailed computational complexity analysis with existing methods.
Response 5: Thanks for the Reviewer’s kind suggestion. We add the complexity analysis of the proposed method. The changes are made in line 292-305..
Point 6: Paper needs proof reading. It contains many grammatical mistakes.
Response 6: Thanks for the Reviewer’s kind suggestion. We have reviewed the full paper carefully and made changes.
Point 7: Reference need improvement as research development and work of year 2022 is missing in references.
Response 7: Thanks for your suggestion. We have added the work of year 2022. The changes are made on page 1, lines 95-99, references 24-26.
Reviewer 2 Report
Comments, suggestions, and errors:
A closing parenthesis is missing in equation (1);
The authors speak of actions (a), but before defining space A (line 179);
In equations (3) and (4) it is important to define θ and Y;
Remove the word “below”, line 219. Give preference to putting “in sequence”;
In equation (5) the parameter η is not defined;
Define acronyms on the first appearance;
Line 274 (algorithm) – action a of space A;
Line 283 – remove the word “above”;
Lines 287 and 288 have errors. Some lines are starting with lowercase letters;
Line 297 has an error – “more than nodes”¿
Check line 301. Quote No. n¿
Lines 300 - Taking into account the issue of reproducibility of the article, describe the Keras platform a little better;
Lines 300 - Regarding the data, it would be important to say if they will be left for other researchers to work with¿
Write “200 bytes” instead of “200 Byte”;
Table 1 – last line: “1 or 30” or “1 to 30”;
Quote Fig. 2 before it appeared. Hit Fig. 2 with the space after it. Do not start item 5.2 with the figure;
Check the text of line 323;
Line 334 starts with a lowercase letter;
Quote Fig. 3 before it appeared. Improve the placement of figure titles;
Quote Fig. 4 before it appeared.
Line 357: What does the x-axis mean¿ Topology snapshot is the number of nodes. Or is it a random topology?
Line 373: define the RSU acronym;
Table 2: Although their references have been cited, the acronyms TDRRL, RAVR, and Q-LBR have to be defined on the first appearance;
Line 388: Set the GPSR acronym
Table 3 is misspelled as Table 2 (line 388) ;
Improve the Conclusions and/or place an item Discussions: If the objective is to send much more types III than types I and II, the system may be worse than the studies used for comparison;
A suggestion of putting something like: “Even when it is not better in all aspects, it can be verified that it is globally better”.
Author Response
The authors would like to thank the Editor and the Reviewers for their precious time and constructive comments and suggestions. Those comments are all valuable and very helpful for revising our paper, as well as the important guiding significance to our researches. We have carefully addressed all the comments. Following this letter are the Reviewers’ comments in black normal font with our responses in red font. Meanwhile, changes made to the original manuscript are highlighted in the revised one. The revision has been developed in consultation with all coauthors, and each author has given approval to the final form of this revision. Our responses are given in a point-by-point manner below.
Point 1: A closing parenthesis is missing in equation (1).
Response 1: We really appreciate you pointing this out. We added parentheses in equation (1).
Point 2: The authors speak of actions (a), but before defining space A (line 179).
Response 2: Thanks for your suggestion. We defined A before we speak of actions (a). The changes are made on page 5, lines 187, 193, 194.
Point 3: In equations (3) and (4) it is important to define θ and Y.
Response 3: Thanks for your suggestion. We describe the parameters in equations 3 and 4 in more detail. The changes are made on page 6, lines 219-226.
Point 4: Remove the word “below”, line 219. Give preference to putting “in sequence”.
Response 4: Thanks for your suggestion. We have changed from “below” to “in sequence”. The changes are made on page 6, line 233.
Point 5: In equation (5) the parameter η is not defined.
Response 5: Thanks for your suggestion. We added the description of . The changes are made on page 7, lines 262, 263, and table 1.
Point 6: Define acronyms on the first appearance.
Response 6: Thanks for your suggestion. We went over the whole article, and defined acronyms on the first appearance. The changes are made on page 2, lines 57, 58 , 69, 70, 75, 76, 89, 90, 93, 94, and page 13, line 453.
Point 7: Line 274 (algorithm) – action a of space A.
Response 7: Thanks for your suggestion. We defined A before we speak of actions (a). The changes are made on page 5, lines 187, 193, 194.
Point 8: Line 283 – remove the word “above”.
Response 8: Thanks for your suggestion. We remove the word “above”.
Point 9: Lines 287 and 288 have errors. Some lines are starting with lowercase letters.
Response 9: We have went over the paragraph and made some corrections. The changes are made on page 9, lines 318-321.
Point 10: Line 297 has an error – “more than nodes”.
Response 10: We really appreciate you pointing this out. The changes are made on page 9, line 338.
Point 11: Check line 301. Quote No. n.
Response 11: We really appreciate you pointing this out. We went over the paragraph and made some corrections. The changes are made on page 9, lines 333-340.
Point 12: Lines 300 - Taking into account the issue of reproducibility of the article, describe the Keras platform a little better.
Response 12: Thanks for your suggestion. We added the introduction of the Keras platform to make it easier for readers to reproduce the paper. The changes are made on page 10, lines 344, 345.
Point 13: Lines 300 - Regarding the data, it would be important to say if they will be left for other researchers to work with.
Response 13: Thanks for your suggestion. Of course, the data could be left for others researchers to work. If researchers need it, we can provide the data.
Point 14: Write “200 bytes” instead of “200 Byte”.
Response 14: Thanks for your suggestion. We write “200 bytes” instead of “200 Byte”. The changes are made on page 10, line 352, and table 2.
Point 15: Table 1 – last line: “1 or 30” or “1 to 30”.
Response 15: Thanks for your question. It's going to be 1 or 30 in table1. To visualize the message types, we set for type I messages that are both urgent and reliable. represents non-urgent but highly reliable type II messages. represents type III messages that are urgent but have low reliability requirements. So and are either 1 or 30. However, the scheme can be set flexibly depending on the types of message.
Point 16: Quote Fig. 2 before it appeared. Hit Fig. 2 with the space after it. Do not start item 5.2 with the figure.
Response 16: Thanks for your suggestion. We have adjusted the position of Fig.2 in the paper.
Point 17: Check the text of line 323.
Response 17: We really appreciate you pointing this out. We carefully reviewed the relevant content and made changes.
Point 18: Line 334 starts with a lowercase letter.
Response 18: We really appreciate you pointing this out. The changes are made on page 11, lines 376-379.
Point 19: Quote Fig. 3 before it appeared. Improve the placement of figure titles.
Response 19: Thanks for your suggestion. We have adjusted the position of Fig.3 in the paper.
Point 20: Quote Fig. 4 before it appeared.
Response 20: Thanks for your suggestion. We have adjusted the position of Fig.4 in the paper.
Point 21: Line 357: What does the x-axis mean¿ Topology snapshot is the number of nodes. Or is it a random topology?
Response 21: Thanks for your question. It’s a random topology. The x-axis in Fig.4 is the topology snapshot number .
Point 22: Line 373: define the RSU acronym.
Response 22: Thanks for your suggestion. The changes are made on page 2, line 63.
Point 23: Table 2: Although their references have been cited, the acronyms TDRRL, RAVR, and Q-LBR have to be defined on the first appearance.
Response 23: Thanks for your suggestion. We added the definition in the paper. The changes are made on page 2, lines 57, 58 , 69, 70, 75, 76, 89, 90, 93, 94.
Point 24: Line 388: Set the GPSR acronym.
Response 24: Thanks for your suggestion. The changes are made on page13, line 453.
Point 25: Table 3 is misspelled as Table 2 (line 388).
Response 25: We really appreciate you pointing this out. The changes are made on page 13, line 424.
Point 26: Improve the Conclusions and/or place an item Discussions: If the objective is to send much more types III than types I and II, the system may be worse than the studies used for comparison;
A suggestion of putting something like: “Even when it is not better in all aspects, it can be verified that it is globally better”.
Response 26: Thanks for your suggestion. We completely agree with the comment made by the Reviewer and really appreciate for pointing this out. We added three message types under the contrast method for performance comparison, and absorbed your suggestions. We have revised the conclusion, which we hope you could accept. The changes are made in lines 409-416, figure 4, lines 459-467, table 4, lines 471-480.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
Paper is well organized. Result section can be improve.
Highlight contribution in bullet points.
Improve conclusion section.
Author Response
The authors would like to thank the Editor and the Reviewers for their precious time and constructive comments and suggestions. Those comments are all valuable and very helpful for revising our paper, as well as the important guiding significance to our researches. We have carefully addressed all the comments. Following this letter are the Reviewers’ comments in black normal font with our responses in red font. Meanwhile, changes made to the original manuscript are highlighted in the revised one. The revision has been developed in consultation with all coauthors, and each author has given approval to the final form of this revision. Our responses are given in a point-by-point manner below.
Point 1: Paper is well organized. Result section can be improve.
Response 1: Thanks for the Reviewer’s kind suggestion. We completely agree with the comment made by the Reviewer and really appreciate for pointing this out. We further improve the full paper. In particular, the result section increases the complexity analysis. The performance is compared with Q-LBR method. The changes are made in lines 324-332, equation 7, lines 409-416, figure 4, lines 459-467, and table 4.
Point 2: Highlight contribution in bullet points. Improve conclusion section.
Response 2: Thanks for the Reviewer’s kind suggestion. We completely agree with the comment made by the Reviewer and really appreciate for pointing this out. We have revised the conclusion in the paper.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
proof reading is needed for some minor english corrections.
Equation format has problems. need corrections in equation format.
Author Response
The authors would like to thank the Editor and the Reviewers for their precious time and constructive comments and suggestions. Those comments are all valuable and very helpful for revising our paper, as well as the important guiding significance to our researches. We have carefully addressed all the comments. Following this letter are the Reviewers’ comments in black normal font with our responses in red font. Meanwhile, changes made to the original manuscript are highlighted in the revised one. The revision has been developed in consultation with all coauthors, and each author has given approval to the final form of this revision. Our responses are given in a point-by-point manner below.
Point 1: proof reading is needed for some minor english corrections.
Response 1: Thanks for the Reviewer’s kind suggestion. We have checked the paper again and made some changes to the syntax and formatting. The changes are made on lines 115, 116, 169, 170, 187, 205-214, 250, 365, 366, 369, 370, and so on.
Point 2: Equation format has problems. need corrections in equation format.
Response 2: Thanks for the Reviewer’s kind suggestion. We have checked the templates carefully. The format of equation 1-7 in this paper is modified.
