MIMO Feedback ( No. of 2D arrays of Channel Matrix, not equal with Precoding Matrix)

[George Frangulea]

Hello,

I tested the beamforming method implemented in the cttc-nr-mimo-demo with the NR-U module. 

The beamforming setup works great for both NR-U/NR-U and NR-U/WiFi. However, when MIMO feedback is enabled, I get an assert in "CalcBeamformingGain" due to the different sizes of precoding and channel matrix. 
From CalcBeamformingGain, I get:
In NR-U/NR-U case the no. of 2D arrays have the following sizes:

• ChannelMatrix numPages: 106 PrecodingMatrix numPages: 106

When in NR-U/WiFi case, sometimes the following no. of 2D arrays come in the function:

• ChannelMatrix numPages: 769 PrecodingMatrix numPages: 106

Does anyone know why this could happen? 
Thank you.

[George Frangulea]

I see this issue happens due to the number of values from SpectrumValue, which for NR SCS 15 kHz is 106, while for Wi-Fi, it comes at 769.

Is there a way to handle the case where there are signals on the channel of different numerologies that will create this issue? Or is this a current constraint when using the MIMO feedback?
Thank you.

[Gabriel Ferreira]

I guess we could downsample/upsample the number of pages to match the precoding matrix.
But we don't have that yet, and you were the first to raise the issue. Please open an issue on the tracker.
And if you have the patches for NR-U (I think you have an open MR), let's get it merged, so that I can help fixing these issues. :)

[George Frangulea]

I think users who use NR only will also find this issue at some point. gNBs can be set with different subcarrier spacing, leading to different number of pages. So, a fix for this would be really good. 

Regarding NR-U, I have been working for 2 years with this module, and the list of merge requests is long. I am happy to work with you on those, but at a later date. Probably from December onwards because now I am trying to finalise my PhD  :D