COPP-DDPG: Computation Offloading with Privacy Preservation in a Vehicular Edge Network
Round 1
Reviewer 1 Report
The problem that this paper addresses is important and overall I like the solution and the presentation. However, the privacy component of the architecture seems to depend on a CA. Some vague statements were made about having to change the certificate regularly but there were no details on this. Also, it wasn't clear how the authenticity of a certificate is checked. This part of the paper has to be improved.
Some minor comments:
It is a little confusing that some references are given without square brackets (eg on page 1, line 29 etc). Please correct this.
Some sentences are hard to read (eg. page 3, line 135: "since VECN is characterized by the highly dynamic such as" doesn't look gramatically correct)
Author Response
Dear reviewer,
We are very grateful for your consideration of this manuscript, and we appreciate the detailed, constructive, and invaluable comments from you. We revised the manuscript according to your comments, and we carefully proofread the manuscript to correct the typographical and grammatical errors. The revision was carried out as a result of your comments. In the item-by-item response attached below, your comments are in black font, and our responses are in red font. Please see the attachment.
Thanks so much for giving us this opportunity to revise the manuscript!
We are looking forward to hearing from you!
All authors
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper contains a significant part of simulations and results analysis that reveal the advantages of the proposed approach in comparison with other methods. However, the conclusions of the paper are somehow unstructured; since the simulations in the paper were focused onto aspects like convergence and overall algorithms comparison, it would have been better to organize conclusions with respect to each aspect that was tested during the simulations. By using this approach, the results of each category of tests would have been better summarized. Also, a paragraph where further development directions are presented is missing from the conclusions chapter, please complete this part.
Besides these minor issues, I really appreciate your work.
Author Response
Dear reviewer,
We are very grateful for your consideration of this manuscript, and we appreciate the detailed, constructive and invaluable comments from you. We revised the manuscript according to your comments, and we carefully proof-read the manuscript to correct the typographical and grammatical errors. The revision was carried out as a result of your comments. In the item-by-item response attached below, your comments are in black font, and our responses are in red font. Please see the attachment.
Thanks so much for giving us this opportunity to revise the manuscript!
We are looking forward to hearing from you!
All authors
Author Response File: Author Response.pdf
Reviewer 3 Report
This manuscript considers the problem of joint computation offloading and resource allocation in a vehicular-edge computing system. This problem appears in autonomous driving applications, augmented reality, and online gaming in 5G and beyond networks. The manuscript employs a privacy-preserving computation offloading framework and formulates a Markov decision process for the inherent cooperative optimization problem. Simulation results are conducted to compare the proposed approach with four algorithms. Results demonstrate the superior performance of the proposed approach.
This manuscript addresses an interesting problem, and it has a good contribution. However, the authors must address some concerns before an acceptance decision is recommended. They are outlined in the following points:
1) Do not use the abbreviation DRL in the title. Please use the whole words and try to make the title shorter.
2) Use the correct reference citation style within the text by embracing the reference number between two square brackets.
3) In line 82, the first sentence of item 3 is incomplete!
4) The scenario described in Section 3.1 needs to be revised for more clarity. It is recommended to have a table of notations and their meanings to facilitate reading the description.
5) In line 161, it is unclear how intelligent connected vehicles (ICVs) follow truncated Gaussian distribution. Such a probability distribution function can be used to model a quantitative measure!
6) In line 162, it is written that the distance between ICVs follows the exponential distribution. What is the mean value that corresponds to that distribution? What is its measuring unit?
7) In the mathematical model (23), is F, which defines the upper bounds of the constraints, the same as italicF? Using the same letter for representing two different things can lead to confusion!
8) In line 345, what is DQN?
9) On page 12, Algorithm 1 needs to be rewritten for the sake of clarity.
10) In line 405, remove “are”
11) It is recommended to separate section 6.3 into two sections. The first one is for the performance of the developed algorithm COPP-DDPG at different algorithm parameters (as in Figures 3 to 5). The other section would be for comparing with other algorithms.
12) The manuscript does not provide an analysis of the mixed effects of the investigated COPP-DDPG parameters on its performance. It is essential to see how the algorithm behaves at different mixed levels of these parameters.
13) The conclusion section is very short. Please provide a summary of the studied problem and its importance, the proposed approach, and the results. Then, provide directions for future research.
Author Response
Dear reviewer,
We are very grateful for your consideration of this manuscript, and we appreciate the detailed, constructive and invaluable comments from you. We revised the manuscript according to your comments, and we carefully proof-read the manuscript to correct the typographical and grammatical errors. The revision was carried out as a result of your comments. In the item-by-item response attached below, your comments are in black font, and our responses are in red font. Please see the attachment.
Thanks so much for giving us this opportunity to revise the manuscript!
We are looking forward to hearing from you!
All authors
Author Response File: Author Response.pdf
Reviewer 4 Report
This is an interesting study. The paper is generally well written and structured. However, in my opinion, the paper has some shortcomings, as provided below.
1. The author discusses privacy preservation techniques as one of the key parts of this work. This part needs more detailed explanation. The author should explain why the proposed method is appropriate for their needs. What are the advantages of using this kind of method for privacy preservation? A detailed comparison of the proposed method and other privacy preservation methods.
2. The author stated in line 200 that “The offloading ratio ??（0≤ ?? ≤1）is depicted as the offloading part ratio to the entire vehicular task ??. ICV-i computes λ_i D_i^in locally and offloads the rest (1 − ??) D_i^in” But at line 475, the author concluded the “the ratio for tasks offloading as ?? = 0.2 in this work.”
If the offloading ratio is set to 0.2, no offloading strategy is required. That implies in equation (23) X is no longer a control variable. This part needs a clearer and more detailed explanation of why all tasks from different vehicles should have the same offloading ratio.
Author Response
Dear reviewer,
We are very grateful for your consideration of this manuscript, and we appreciate the detailed, constructive, and invaluable comments from you. We revised the manuscript according to your comments, and we carefully proofread the manuscript to correct the typographical and grammatical errors. The revision was carried out as a result of your comments. In the item-by-item response attached below, your comments are in black font, and our responses are in red font. Please see the attachment.
Thanks so much for giving us this opportunity to revise the manuscript!
We are looking forward to hearing from you!
All authors
Author Response File: Author Response.pdf
Round 2
Reviewer 4 Report
No further comment.
