kumoh national institute of technology
Networked Systems Lab.

Ade, "CNN-Based Automatic Modulation Classification for Beyond 5G Communications", IEEE Communications Letters (Minor Revision)
By : Ade Pitra
Date : 2020-01-12
Views : 279

Dear Author(s):

The review of the referenced manuscript CL2019-2714, entitled CNN-Based Automatic Modulation Classification for Beyond 5G Communications is now complete. The reviews of the manuscript are attached. Based on the reviews and my own reading of your manuscript, I cannot accept your letter for publication in its current form. Your manuscript requires revisions, as outlined below, before the paper can be published. Two of the reviewers still have some concerns and valuable suggestions, such as: 1) the motivation for the AMC problem and the use of DNN must be emphasized; 2)a fair comparision in term of structure and key defferences to that of [6] should be included. Please carefully revised the manuscript according to the comments provided by the reviewers and if these revisions are satisfactorily made (including meeting the length guidelines), the paper will be accepted for publication.

Your revision will be due within 21 days and is due on 02-Feb-2020. Please ensure that your revision is submitted in a timely manner as the web-based system will not allow a revision to enter the system after 21 days have elapsed.

Thank you for considering IEEE Communications Letters as a means of publication. I look forward to receiving your revised manuscript.

Best Regards,

Dr. Animesh Yadav
Associate Editor
IEEE Communications Letters


Reviewer's Comments

Reviewer: 1

Comments to the Author
Most of the comments have been addressed and the domain of application has been changed to B5G. Some minor changes could make the paper more robust in terms of their comparisons conducted in this work. There are also parts of the writing that can be improved/corrected. Once these changes are made, the paper should be ready for acceptance. Please see the attached document.

Reviewer: 2

Comments to the Author
No comments.

Reviewer: 3

Comments to the Author
The authors propose to solve the AMC problem using the DNN-based approach and have provided some exciting results. I have the following comments:
1. The authors need to explain how they manage to improve computational time. They have mentioned that they 'apply an additional number of filters' and have still managed to reduce the processing time, which is counter-intuitive.
2. In the Performance Evaluation section, the lines 'Fig. 4 shows that the accuracy rate of our IC-AMCNet still outperforms the other algorithms. The highest accuracy of IC-AMCNet 0.4 dropout is around 81.43% at SNR 18dB, where for CNN1, CLDNN, CNN2, NN2 hidden layers, and IC-AMCNet 0.7 dropout is 78.75%, 82.28%, 80.49%, 59.85%, and 82.89% respectively.' are contradicting. CLDNN has better accuracy than IC-AMCNet. Further, based on the plot, it is hard to see the proposed method is performing better than other algorithms.
3. In the first two contributions, the authors tell they are increasing and reducing the number of filters. Please be clear on the contributions.Dr.
Attachments
Attachment 1:   ÷ Ade IEEECOML Rev2.pdf(104.0KB)