A Light Signaling Approach to Node Grouping for Massive MIMO IoT Networks

Round 1

Reviewer 1 Report

In this paper, the authors propose light signaling to node grouping and compare it to the Clumped, power, and optimal methods. The procedure used in this paper is acceptable. In order to find a high-level paper to be cited by other researchers, the following items should be considered.

1.       It is proposed and tested for a system model consisting of one base station. How about the cases where there are more than one base station or multi-layer scenarios in 5G?

2.       The light signaling tries to do the optimal solution in an approximated form. I do not know why Equal gain combining (EGC) has not been used instead of MRC?

3.       In this paper the details of the channel model have not been detailed. The noise model, large-scale and small-scale phenomena such as multipath and shadowing ones, as well as a path-loss model, should be detailed.

4.       The effect of incomplete CSI has not been evaluated.

5.       In addition to SINR, in performance evaluations and comparisons, outage probability and ergodic capacity or achievable rate should be extracted.

6.       To have a fair comparison, time and computational complexity should be analyzed.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors,

Congratulations for your paper. I would like to provide you a few comments in order to improve the quality of the paper.

- It is recommended that if massive machine-type communication uses the acronym mMTC, then massive multiple-input multiple-output should use mMIMO.

- When in row 522 is mention N, it should repeat the value in row 542.

- It is highly recommended to review the English grammar. Some sentences are difficult to understand. i.e. One approach that has similarities to ours, albeit not aimed at node partitioning for scheduling, is that of joint spatial division and multiplexing.

Kind regards,

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors propose for massive MIMO IoT networks node partitioning strategies that do not require full channel state information, but rather are based on nodes’ respective directional channel properties. The paper is interesting. I have the following comments and questions.

-In the abstract it is written "In our considered scenarios, these typically have a time constant that is far larger than the coherence time of the channel". If your IoT devices are static their velocity is null and the coherence time of the channel tends to infinity. What is not clear in your scenario is how many IoT devices are communicating with the acess point (AP)? 

- On Table 2 is presented the parameters for the simulation. For 36 nodes and 100 antennas there is less than 3 antennas per user. Is this ratio enough for the IoT devices?

-Table 2 has SNR=20dB for all nodes. Are they all at the same distance from the AP?

-Table 2 has 12 pilots. Why not choosing 36 pilots equals to the maximum number of nodes?  Are there no 36 pilots with good cross-correlation properties? 

-Table 2 has fc=2.47GHz, so I assume that you are considering WiFi. You are considering Indoor scenarios. What is the area of your simulation scenario, 500 square meters?

- here are two papers refering to the simulation indoor scenario, one of the is from 1987. The other is for mmWve MIMO channels. So, it is not valid for 2.47GHz.

- Your simulation results are quite short. Basically only two figures, namely, Figure 4 and 5. You should consider other fc different from 2.47GHz and provide new simulation results for this new carrier frequency fc. I'd suggest the 5GHz band for WiFi with at least 200 antennas at the AP.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Most of the comments are addressed carefully. The current version of the paper is appropriate for readers to follow it.

Author Response

We thank the reviewer again for his or her time and feedback.

Reviewer 3 Report

The authors say that "A light signalling user grouping approach as we have demonstrated in this paper is valuable for any situation where signalling overhead is significant. This will be the case when there are a very large number of nodes, as in IoT scenarios." However the simulations results presented in this paper only considered 36 nodes (IoT devices). This is not a real number of IoT devices for many cases. As a consequence your results are not in agreement with your statements.

Author Response

We have added the following sentence in the conclusion of the paper acknowledging this point:

"Although we have thus far only tested cases with up to 36 nodes, and in real IoT deployments the number of nodes is expected to be much greater, the approximation algorithm we have presented shows good scalability and should thus be suitable even in larger scenarios."

We agree that larger scenarios should be tested in order to be more in line with the expected size of many IoT deployments, however in this initial work we were limited by the scalability of the optimisation algorithm. In future work we plan to focus on the approximation algorithm and can then test larger problem instances.