Tool/software:
Hi TI experts,
I am working with the AWR2944EVM OOB and HECR demos, and my goal is to extract from the raw radar cube a slice of post-range FFT data at a given range bin, across all chirps and all antennas. I have described the overall idea in more depth here: AWR2944EVM: Acces and compression of radar cube data
Since the DDMA demos use compression, I currently copy the data from the decomposition scratch buffer in dopplerprochwaDDMA.c. However, I suspect that I may be reading the wrong data, and I am not sure why, so I am reviewing the key assumptions and workings of the code.
According to the documentation:
1.) Inside dopplerDDMA, section "Windowing and Doppler FFT":
* The EDMA source is the decompressed cube scratch buffer. The documented format is: [numChirps][rangeBinsPerBlock (indexed via rangeBinIdx)][rxAntennas]
* file:///C:/ti/mmwave_mcuplus_sdk_04_07_00_01/mmwave_mcuplus_sdk_04_07_00_01/ti/datapath/dpu/dopplerprocDDMA/docs/doxygen/html/index.html
2.) Inside rangeprocDDMA, the internal compression dimension inside an “EGE” block isn’t clearly specified. Is the compressed [rangeBin][rx] format or the other way around? The most I could find is: "in case of EGE compression, each Block consists of compressed range bins per block * RX antenna samples." But it is not clearly specified what is the in memory layout.
* file:///C:/ti/mmwave_mcuplus_sdk_04_07_00_01/mmwave_mcuplus_sdk_04_07_00_01/ti/datapath/dpu/rangeprocDDMA/docs/doxygen/html/dpu_rangehwa_d_d_m_a.html
3.) Moreover the in memory layout of the data seems to be fundamentally different in case of BFP compression.
My questions:
I.) What is the actual layout inside the scratch buffer?
a.) Does the layout change depending on the compression method used (EGE vs BFP)?
b.) Or is the decompressed buffer always of the form: [numChirps][rangeBinsPerBlock (indexed via rangeBinIdx)][rxAntennas] regardless of the compressed memory layout?
Thank you for your support
Mark
