Controllable Wireless Spoofing Attack Based on Conditional BEGAN and Auxiliary Channel Sensing

Round 1

Reviewer 1 Report

Recommendations

In the article, “Controllable Wirelss Spoofing Attack Based on Conditional BEGAN and Auxiliary Channel Sensing." the paper proposes a wireless spoofing attack scheme against the defense mechanism with adversarial deep learning. The authors must improve the manuscript. However, I have some comments and suggestions for the authors. as follows:

 

1-  The novelty of the paper is not clear. To be more specific, optimizing communication and computation resources have received much attention in the literature. Still, the paper did not state what the proposed approach could add compared to the state of the art.

2-  The authors must improve the system model because the system lacks the mechanism of spoofing prediction for the LOS path signal and powers of the remaining multipath.

3-  The equations in a system model must support with citation references.

4-  The section for results on spoofing is not related to the system model in terms of analysis results. So, the author must improve the system model and make it more related to Simulation results.

5-  Good analysis of the proposed method, simulations were performed via multiple aspects. However, the baselines were not convincing enough. Apart from the general models of GAN-based attacks, the result is only compared with a proposed attack. It is encouraged to include some more recent works as the baseline.

6-  It seems that the GAN-based attack can also perform well in the simulation result, with a simpler optimization. So, what is the main advantage of the proposed work compared with these works?

 

 

Major correction 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper advises a manageable wireless spoofing attack scheme against the 335 deep learning-based classifier with AAE-based CBEGAN and supplementary channel sensing.  Using the training a CBEGAN with the joint circumstances including channel sensing characteristics and embedding signal labels, the spoofing attack might be launched controllable. was introduced in the architecture of CBEGAN to train the networks with few samples. Experiments were conducted using USRPs to capture wireless signal and propagation data for a multi-emitter wireless scenario. Untried results verified the advantages of the proposed scheme over the random attack, replay attack, and GAN-based attack in the situation where a single emitter sent signals with only one modulation type. Results shown that the suggested scheme could mimic different emitters with different modulation types.

This paper advises a manageable wireless spoofing attack scheme against the deep learning-based classifier with AAE-based CBEGAN and supplementary channel sensing.  Using the training a CBEGAN with the joint circumstances including channel sensing characteristics and embedding signal labels, the spoofing attack might be launched controllable. was introduced in the architecture of CBEGAN to train the networks with few samples. Experiments were conducted using USRPs to capture wireless signal and propagation data for a multi-emitter wireless scenario. Untried results verified the advantages of the proposed scheme over the random attack, replay attack, and GAN-based attack in the situation where a single emitter sent signals with only one modulation type. Results shown that the suggested scheme could mimic different emitters with different modulation types.

1.    The AAE, why?

2.    how do you choose the training hyperparameters of CBEGAN.

3.    What about a real setup of measurements?

4.    Some explanation at level of the performances must be discussed deeply?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed all of the comments. No further comments

Reviewer 3 Report

The updated manuscript looks good.