An Artificial Intelligence-Assisted Expectation Propagation Detection for MIMO Systems

Round 1

Reviewer 1 Report

This is a well-written manuscript on a new expectation propagation detection algorithm for MIMO systems. It will interest a reasonably broad readership of Electronics. I have the following comments that the authors should address before final acceptance:

1.     In table 1, the authors listed several MIMO detection algorithms, as well as the basic descriptions. It would be better to also list the complexity of each algorithm. It would be more helpful to quantify the performance of different algorithms in this table.

2.     For figures 2, 3, 4, and 5, it would be better to plot similar colors for EP and MLP with the same iteration numbers. For example, red for 2 iterations, green for 3 iterations, etc., so that it is easier for readers to compare. (currently 2 iterations(EP) and 4 iterations(MLP) have similar colors, which is a bit hard to follow and compare.)

 

3.     ML detector is an NP-hard problem, but there are lots of literatures transformed this problem into an Ising minimization problem and efficiently solved it using an Ising Machine. Can the algorithm proposed in this manuscript outperform them?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper discusses artificial intelligence-assisted expectation propagation detection for MIMO systems. The paper can be improved if the following points are considered.

 1. It is advised to remove equation from the introduction section and move it to another appropriate place. However, if authors feel it is absolutely necessary, then it can be kept. Also, in (1), several symbols are not defined.

2. Diversity in writing  for line 81-83 is required.

3.  The contributions of the paper are not well defined.

4. Organization of the paper should be included in the end of section 1.

5. Which method authors have considered for defining vectors and matrices is not clear and shall be included in the paper.

6. Line 91-92 does not define well the subsequent expression. Also, where in the effect of mentioned interference.

7. Please define the limits of integration in (4).

8. Section 2 ends without stating how the system model will be considered later and what analysis/investigations are to follow.

9. SNR range in all figures shall be increased to accommodate more analysis data. X-axis shall be labeled as dB scale if that is the case.

10. Statistical analysis of BER for the schemes is missing in the results section. For such, box plots of BER for the BER can be used as in

11. Line 336-354 is about computational complexity, and it is better to include the same as a Table in results section.

12. Some of the recent relevant MIMO channel schemes may be cited e.g. https://doi.org/10.3390/s21206792

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Please create performance comparison table with the other published articles.

How about mAP(%) and operation time? Please comparison with propagation detection for MIMO systems

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors have addressed all the points made in my first review report. I think it is acceptable for publication in the journal.

 

Reviewer 2 Report

Authors have addressed my comments well.