is there a mistake in "Deep Learning Data Synthesis for 5G Channel Estimation" example code?

In the "Compare and Visualize Various Channel Estimations" part of the example code, it used nrChannelEstimate function to get the practical CE. However, it's using the rxGrid which takes an offset provided by nrPerfectTimingEstimate. Could someone help me to confirm this? Thank you!

The link to the code: Deep Learning Data Synthesis for 5G Channel Estimation - MATLAB & Simulink - MathWorks Nordic

Answer 1

[Aman]

Hi,

I understand that you are trying the "Deep Learning Data Synthesis for 5G Channel Estimation" example and are having trouble in understanding why offset from "nrPerfectTimingEstimate" is used for getting the resource grid (“rxGrid”).

The example talks about how the channel estimation task can be performed using deep learning, and for that, the resource grid is considered a 2D image, and then a CNN model is made for the channel estimation.

The discussed approach is also compared with perfect channel estimation and practical channel estimation. "nrChannelEstimate" is used to get the practical channel estimation; this method needs a resource grid as an input, which is "rxGrid" in the example. To get the “rxGrid", the offset from "nrPerfectTimingEstimate" is used, as it is assumed that there is no timing error.

In case you want to use the practical timing estimate, please refer to the following documentation:

https://www.mathworks.com/help/5g/ref/nrchannelestimate.html

I hope it helps!