Next Article in Journal
Revealing IoT Cryptographic Settings through Electromagnetic Side-Channel Analysis
Next Article in Special Issue
Computation Offloading Strategy for Detection Task in Railway IoT with Integrated Sensing, Storage, and Computing
Previous Article in Journal
Improved Fault Diagnosis of Roller Bearings Using an Equal-Angle Integer-Period Array Convolutional Neural Network
Previous Article in Special Issue
Design of a Collaborative Vehicle Formation Control Simulation Test System
 
 
Article
Peer-Review Record

Intelligent Security Authentication for Connected and Autonomous Vehicles: Attacks and Defenses

Electronics 2024, 13(8), 1577; https://doi.org/10.3390/electronics13081577
by Xiaoying Qiu 1,*, Jinwei Yu 2, Wenbao Jiang 1 and Xuan Sun 1
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Reviewer 4: Anonymous
Electronics 2024, 13(8), 1577; https://doi.org/10.3390/electronics13081577
Submission received: 8 March 2024 / Revised: 8 April 2024 / Accepted: 18 April 2024 / Published: 20 April 2024
(This article belongs to the Special Issue Control Systems Design for Connected and Autonomous Vehicles)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript investigates a GAN-assisted authentication framework for connected autonomous vehicles. However, my main concern is the presentation of the work. The reviewer can not grasp what are the main idea of the paper or the main contributions compared to the state of the art.  The following are some other issues:

1- What is the meaning of Figure 2 and its relation to the idea of the paper? It is not explained well. 

2- The equation number is wrong in line 132

3- add a reference for the first paragraph of sub-section 3.2

4-I line 172, the author claimed to introduce two constructions called NRCT and 4CRd,  however, this abbreviation has not been introduced before nor has a figure showing its architecture

5- In line 222, the authors mentioned two CSV files but no reference or proper description 

6- In line 230, two variables are described to use the same symbol. correct that

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The paper presented the Generative Adversarial Network Learning-Assisted Authentication Framework for Connected and Autonomous Vehicles: Attacks and Defenses.

However, I have some comments as follows:

1. In the abstract, please elaborate on the proposed system for better reader understanding.

2. It is suggested that the motivation be clarified, and a related work section with GAN-related studies should be added.

3. In this paper, the authors utilized NIST datasets. However, the dataset preparation and overall description need to be addressed. Additionally, comparing the simulation results with other datasets within the same framework is recommended.

4. In the simulation, please provide a comparison of previous related studies with the proposed system.

5. The reviewer is curious about the model and how the authors determined that the presented results are the best outcome. Provide a strong justification with results by tuning the hyperparameters of the models.

6. It is highly recommended to compare the complexity of the proposed scheme with others.

 

Comments on the Quality of English Language

Moderate editing of the English language required

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The paper proposes a generative adversarial network architecture without pooling layers to detect security attacks in noisy wireless environments for autonomous vehicles. The motivation of the paper is good but there are lots of concerns with the work. The authors are encouraged to address the following concerns before this work can be publishable.

1. The abstract is not detailed enough. Signal enhancement and its importance are not discussed. The benefits of GAF over other methods are also not highlighted.

2. The importance of Equations 1 and 2 in the paper is unknown. If they are not important to the overall idea presented by the authors, then they should be removed. 

3. Some statements seem copied from a different source and have no clear connection with the paper, thus distorting the logical flow of the paper. For instance, the authors wrote "The effect of the fake data set on the authentication model mainly depends on the number of malicious vehicles involved k and the proportion of the respective amount of tampered data ai. According to Eq. (2.1), when k << I or the amount of tampered data ai is small, the spoofing attack has the least impact on the authentication model, but with the increase of the number of pseudo-attack vehicles, the pollution degree of the authentication model will be greatly deepened."  Where is Eq.(2.1) in the paper? Where is k? 

3. The authors claimed that the training set was divided into training and test sets. Was it the original dataset that was divided or the training set? What is the size of this dataset? More so, the authors claimed that the division was 90% for training and 10% for testing in Section 2.2 but later wrote that they used 90% for training, 5% for verification, and 5% for testing in Section 4.1. So, why the inconsistencies? what is verification, is it the same as validation?

4. The authors wrote "The NRCT is internal structure of the generator and includes the normalization module, ReLU activation, convolutional layer, and Tanh activation. The 4CRD is the internal structure of the generator, consisting of four convolutional layers, ReLU activation, and Dropout layers".  Is NRCT and 4CRD both internal structures of the generator? How about the discriminator?

5. The authors claim that they remove white noise from the signal. It is unclear to me how white noise can be eliminated from a wireless signal. Can the authors provide supporting evidence?

6. The authors claimed that the noise signal gradually converged. Can noise signals converge? This is unrealistic.

7. The authors compare their work with MMSE-SPZC and SEGAN without referencing papers that proposed these algorithms.

7. The authors used some metrics such as PESQ value, average STOI value, and SNRseg value, but did not define these metrics.

8. This statement cannot be deduced from Table 3, "The signal-to-noise ratio of MMSE-SPZC algorithm is low at -2.5dB and 2.5dB. Therefore, it can be concluded that the MMSE-SPZC algorithm has basically lost the ability to process noise signals and remove noise. In addition, from the last two columns of the table, it can be seen that the SNRseg values of GAF algorithm and SEGAN algorithm are lower than the original SNRseg values of noisy wireless signals, and some of them are even lower than the initial SNRseg values of noisy wireless signals."

9. The below statement is incorrect because, at iteration 420, there is still fluctuation of R2 score, TSS, RSS, and ESS

"With the increase of training data, although the R2 score, TSS, RSS, and ESS function values fluctuate, they remain stable after 420 iterations."

10. There are lots of typos in this work and careful revision is necessary.

Comments on the Quality of English Language

Revision must be done to improve the English Language.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

The article technically sounds, and the main contributions are clearly introduced. I have the following minor concerns.

-        The work requires extensive proofreading; there are many typos and grammatical errors.

-        The introduction section is too long.

-        Avoid using long paragraphs.

-        Many parts are overcited; unnecessary citations are used. The authors should consider relevant recent citations in the necessary places. Also, some parts miss citations.

Comments on the Quality of English Language

he work requires extensive proofreading; there are many typos and grammatical errors.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed my concerns. I have no further comments

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have addressed all of the comments properly.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have addressed all my comments; I have no more comments. The work can be published in the current form.

Comments on the Quality of English Language

Minor editing is required.

Back to TopTop