Signal Modulation Recognition Algorithm Based on Improved Spatiotemporal Multi-Channel Network

Round 1

Reviewer 1 Report

Based on the already-existing modulation signal identification algorithms, the authors of the paper "Signal Modulation Recognition Algorithm Based on Improved Spatiotemporal Multi-Channel Network" suggested an improved spatiotemporal multi-channel network model.

 

Concerns

• Lack of sufficient detail prevents readers from completely understanding and replicating the authors' analyses and experiments. 
• Lack of sufficient detail prevents readers from completely understanding and replicating the authors' analyses and experiments.  
• The paper has poor language quality, typos, long sentences without punctuation, and acronimis without being previously addressed. 
• Features prepocessing, features dataset organization, labeling, and deep models lack meaningful details. 
• The paper has difficult-to-follow logic and poorly presented data. 
• Train and test validation methods lacks sufficient details. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper works on signal modulation recognition, the method is well designed and the experimental results seem good. The paper should be improved from following aspects:

1. Figure 3 shows the architecture of LSTM, which is well known to all. I don't think it is necessary to put this figure in the paper.

2. The key novelties and contributions of the paper are not well highlighted.

3. The reference section is weak, only 20 papers are not sufficient to review the existing works. More related works should be added, including but not limited to:

a. Seismic envelope inversion and modulation signal model, Geophysics 2014.

b. Multiview and multimodal pervasive indoor localization, ACM Multimedia 2017.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

• The goal is poorly handled, and the background is weak. 
• The presentation of the methodology and outcomes is unclear. 
• There has to be extensive English checking.
•  
•  

  1.      Lines 16 through 18 The authors write: "In this study, we suggested an enhanced spatiotemporal multi-channel network to address this issue based on the existing MCLDNN method (IQ-related features Multi-channel Convolutional BiLSTM with GaussianNoise, IQGMCL). However, it is unclear how the MCLDNN method is connected to the IQ-related features Multi-channel Convolutional BiLSTM with GaussianNoise, IQGMCL.

  2.      I would advise including a sentence on signal modulation in the Introduction.

  3.      What does corrosion mean in lines 97-98?

  4.      The reader can inquire after reading the introduction, "What are IQ signals?"

  5.      I would suggest references in table 1

  6.      Figure 1: Plots lack x-y labels, the caption is weak  and the differences between the classes are not obvious. Additionally, the issue of the space/time signal, which is unclear throughout the text, may be addressed here.

  7.      I'd recommend information on signal preprocessing and dataset organization, shape, and size.

  8.      Figures 3 and 4 should be more specific; they might represent an informative architecture with input size, layers, and hyperparameters to fit with those used in this study.

  9.      A reference in lines 211-213 would be helpful.

  10.    Lines 218–220 are ambiguous, and it is unclear how the signal has been merged.

  11.    The class labels and dataset labeling process are unclear.

  12.    There was no mention of the train-test-validate method.

  13.    Because the process for doing various analyses based on SNR adjustment is unclear, Figure 6 is also unclear.

  14.    What does Mcc and AACC mean?

  15.    Lines 294–297 introduce multiple deep models. However, the reader is still naive to the characteristics of the architecture and hyperparameters, though.
•  

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

My concerns are addressed. I recommend to accept this paper.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

The authors did an excellent job of improving the overall sound during the revision. I have no further feedback to offer.