Problem in the I/Q HDL test bench data generated from WLAN HDL transmitter/receiver

Dear all,

I am using these two examples TX and RX to generate HDL test bench with test data.

After opening the module in matlab, I think the configuration for mcs and frameformat can be changed, then we can re-generate the test bench and test data with new configurations.

However, I found that the I/Q test data look strange.

TX default mcs is 0 (BPSK), so there should be 2 clusters at the I/Q constellation, and not imag part.

RX default mcs is 7 (64-QAM), so there should be 64 clusters at the I/Q constellation.

Taking TX for example: somehow I found there are both real and imag data (should be no imag data for BPSK, I think).

The real data look like: 006836bf, fed331c2, ffe11c73, 01447a3c, ......, so either very large or close to 0.

The HDL code specifies sfix32 en24 for the I/Q data.

Then, I try to make scattering plot for different interpretations. I have no idea which one is the correct one:

Next, taking RX as example, where the default is mcs = 7 (64-QAM).

The HDL code specifies sfix16 en12 for the I/Q data.

Then I did the same plots, and they also look strange:

All the plots above are based on golden I/Q data generated by matlab.

Could anyone please help to clarify:

• After changing the mcs and frameformat setup in matlab/simulink module, is the re-generated HDL test data also changed?
• How could I correctly interpret those generated I/Q data from TX and RX examples?

Thank you in advance.

Answer 1

[Yun-Tsung]

I found the reason. Those I/Q data are in time-domain, so the scattering plot doesn't give the constellation information.

We have to use the data without OFDM modulation, then the scattering plot will show the points with constellation information.

[Kranti Balaga]

Yes correct, you can tap after symbol modulator subsystem. I have checked for 64-QAM, you can get the following actual constellation.