Problem in Frequency domain of RADAR GUI

[fpga]

Hello again,

Anybody can explain why at 2,4 m, I got the returned signal with Magnitude = 62.18 dB, but at 1.6 m I just got 55.54 dB!!

Your answering my question would be appreciated.

Best regards!

[Honey_D]

Hello @fpga ,

The FFT Magnitude view shows FFT magnitude value of the strongest target detected by a device. D2GL algorithm is for detection and tracking of human or vibrating objects.

On each chirp per frame received, the range fast Fourier transform (FFT) is performed. In the next steps, only the range bins between minimum and maximum range are investigated. The magnitudes of the range FFT data are not used to detect and track the targets, but target activation data and target hold data are calculated and utilized. While target activation data is mainly used to detect targets and activate the tracks, target hold data is used to check if activated tracks are still alive.

Please refer to the help content for detail explanation on the Radar GUI parameters and settings. Also to Software manual for algorithm details.

Can you please share the Radar GUI parameter setting snapshots and please confirm the board you are referring to?

Best Regards,

Honey

[fpga]

Hi Honey,

If possible, can you talk more about how the "target activation data" is used for detecting targets? Can you share the source code on how this function is established (may be in JAVA programming language?) to detect the distance from the raw data signal that is acquired from RADAR GUI?

The board and the platforms are BGT24LTR11 and distance2gol. I also add here the snapshots that show the Radar GUI parameter setting. 

Thank you for your support!
Best regards,

[Honey_D]

Hello @fpga ,

Please install the Distance2GoL kit from Infineon developer center for all the required documents and source code required.

Best Regards,

Honey

[Honey_D]

Hello @fpga ,

The Radar GUI shows the frequency domain data based on FFT computation with the GUI itself. The closer the target the higher the peak for same settings. You can refer to the target hold data for clear picture about the target range and magnitude

We would suggest to test the MMIC in outdoor environment without any obstacle as multiple reflection from small objects may also create such ambiguity in the result.

Do let us know about your findings and also if this works for you.

Best Regards,

Honey

[fpga]

Thank @Honey_D for your answer,

 I saw there is a problem with the display (or maybe computation) in the frequency domain.
I have made my measure in an outdoor environment. However, at 1.65 m, I got 55.54 dB and 62.18 at 2.44 m

Furthermore, as I mentioned in another thread:
Solved: Re: BGT24LTR11 with Distance2Gol kit - Infineon Developer Community
Even in your demo video, there is always this problem.
I am looking in your source code at the firmware level, if I can find any useful information, I will come back to you

Thank you for your supporting

[12]

[fpga]

Hi,

Just look into the source code of Matlab interface in /IFX_D2GL-HW-SW_V1.0.0/Firmware_Software/Communication Library/ComLib_Matlab_Interface/Examples/GettingStarted

I saw that:
hSub1 = plot(ArrayBinRange_plot,mxResultData.activation_data);

hSub2 = plot(ArrayBinRange_plot,mxResultData.hold_data);

It means that the activation_data and hold_data are calculated in the Firmware level (DAVE TM application).

Best,

[Honey_D · Accepted solution]

Hello @fpga ,

The frequency spectrum is calculated in the Radar GUI only, it is not used within the algorithm, thus it is just for a comparison to state-of-the-art detection algorithms based on range FFT amplitude. The curve of the spectrum is normal, as in the lower range bins, thus around 2m, we see some larger magnitudes due to the frequency dependent TX-RX leakage as well as the transient due to the DC offset jump at the input of the operational amplifiers in the analog baseband due to the change of the frequency linearity at the start of each chirp. Yes, the results for the algorithm output shown in the Radar GUI are the same as only Target Activation Data and Target Hold Data are utilized within the algorithm.

Data can be recorded in Radar GUI and the stored files can be loaded in the MATLAB parse example, which is delivered with the D2GL package. Here, the raw data can be loaded, amplitude spectrum can be calculated using fft() and abs(), and the target activation and hold data are also available in the stored files (they are actually plotted in this example).

If you are still getting complete different signal curves, then we would need a picture of the measurement setup, to understand what the radar "sees". Also with reference to the image regarding the settings you shared, we would suggest you to use the latest Radar GUI version

Hope this helps

Best Regards,

Honey