IWR1443BOOST: IWR1443BOOST Heatmap

Hi Henry,

Are you working with the standard Out of Box Demo or a different software set?

We'll need a few days to respond to the rest of your question.

Regards,

AG

Hi AG,

I am using the default, preloaded software of the demo board and I am using Python to get data from the comm port using the CLI commands.

Thanks a lot,

Henry

Hi, Henry:

Which SDK version you are using?  And are you talking about range-Doppler heatmap?   If so, the range-Doppler heatmap size is determined by the number of ADC samples (range FFT size) per chirp and the number of chirps (Doppler FFT size) in one frame.  

You can search for profileCfg and frameCfg inside the SDK users guide to understand how to change them in CLI control interface.  The SDK users guide is located at:  C:\ti\mmwave_sdk_xx_xx_xx_xx\docs\mmwave_sdk_user_guide.pdf.

The range of ADC samples is below 1024, which can be found below:

file:///C:/ti/mmwave_sdk_03_03_00_03/packages/ti/control/mmwavelink/docs/doxygen/html/structrl_profile_cfg__t.html#a9448ecc83ce18857f1f1efba82af3d72

The range of number of  loops is below 255 , which can be found below:

file:///C:/ti/mmwave_sdk_03_03_00_03/packages/ti/control/mmwavelink/docs/doxygen/html/structrl_frame_cfg__t.html#a6f3cf525d5792d9f906c72cd7e76f66d

In addition there is limitation of the max value of (range FFT size * Doppler FFT size * number of antenna pair), which is below 96K complex samples per frame for IWR1443.

In addition, you may limited by UART speed, so please adjust the frame periodicity accordingly. 

Best,

Zigang

Hi Zigang,

It is MMWAVE SDK 2.1.0.4 and I am doing the Azimuth heatmap.

Thanks a lot,

Henry

Hi, Henry:

The Range-Azimuth heatmap is generated from the zero Doppler bins.   The zero Doppler bins is calculated inside radar sensor and with size (number of range FFT * number of visual antenna), then the actual heatmap is calculated inside the visualizer instead of the radar sensor. 

I will reach out to find out the size of this heatmap in angle domian and get back to you.

Best,

Zigang

Hi Zigang,

What is the number of range FFT?

Thanks a lot,

Henry

HI, Henry:

Range FFT information can be configured through profileCfg as number of ADC samples.   You can search for profileCfg inside the SDK users guide to understand how to change them in CLI control interface.  The SDK users guide is located at this relative directory:  C:\ti\mmwave_sdk_xx_xx_xx_xx\docs\mmwave_sdk_user_guide.pdf.

The maximum number of ADC samples is below 1024.  

In addition there is limitation of the max value of (range FFT size * Doppler FFT size * number of antenna pair), which is below 96K complex samples per frame for IWR1443.

Best,

Zigang

Hi Zigang,

Actually, I have two more questions:

1. I am trying to do the calculation on the angle of arrival based on the heatmap data. What is the FFT bin to do the AOA? I am using 8 which is the number of virtual antenna. But I looked at the code of mmWave Demo Visualizer, it seems using a bin of 64. 

2.I am using 2 TX and 4 RX antennas for MiMo, would the virtual antenna sequence be in the heatmap data as: virtual antenna 1, virtual antenna 2 .... virtual antenna 8. If we do FFT on that and find the the first peak at sequence number 5, is the object at the azimuth angle of 90 degrees?

Thanks a lot,

Henry

Hi Zigang,

Does 'fftshift' mean doing FFT? If yes, I don't understand why w_peak is -pi. In my example, the index of peak of FFT (Kmax) is 5,  w_peak should be '0' in the FFT output sequence of [-4, -3, -2, -1, 0, 1, 2, 3]. If w_peak is zero, theta should be zero as well.

Please check if my calculation is correct.

Thanks a lot,

Henry

Hi Zigang,

I see. But based on the procedures on the Angle of Arrival calculation suggested by the TI's app note and the reference document 'Millimeter Wave (mmw) Demo for XWR14XX', should we use the output index of  (-4, -3, -2, -1, 0, 1, 2, 3) or  (0, 1, 2, 3, -4, -3, -2, -1)?

Thanks again,

Henry

Hi, Henry:

Can you send me the related code you refer to?  

Best,

Zigang

Hi Zigang,

I have spent some time on checking/correcting the codes and doing some verification tests. Finally, the AOA is working according to your suggestions - need to do the 'fftshift'. But my calculated AOA is not as accurate and stable as the one found from the demo software 'Mmwave Demo Visualizer'. I checked the code of the AOA calculation file "mmWave.js" of the Visualizer software, seems that it only find one single object (the peak one) in each range bin. Am I correct? Also, any references on the detailed calculation of AOA and multiple object detection in the demo software 'Mmwave Demo Visualizer'

Thanks a lot,

Henry

Hi, Henry:

Glad to know that you have made some progress.   When trying to detect multiple peaks from the angle spectrum, you can get false detection due to the sidelobe in the spectrum.   We do not have multiple peak detection in SDK release demo.   If I can find a multiple peak detection example in other TiRex labs, I will let you know.

But you can find some document on AOA at following location.  

file:///C:/ti/mmwave_sdk_03_04_00_03/packages/ti/datapath/dpc/dpu/aoaproc/docs/doxygen/html/index.html

file:///C:/ti/mmwave_sdk_03_04_00_03/packages/ti/datapath/dpc/dpu/aoa2dproc/docs/doxygen/html/index.html

Best,

Zigang