SAF86xx FFT

[Milton]

Hi,

We are evaluating the SAF86xx.

In the description, SAF86xx provides pre-processed radar data, e.g., compressed or uncompressed Range-FFT, over Gigabit Ethernet to a radar post-processing unit.

I'm curious about the meaning of 'compressed' and 'uncompressed' in the context of SAF86xx. Could you please provide some clarification? 

Take absolute value or use smaller bits?

If we want to do further signal processing, for instance generate point clouds, is it available to generate it by compressed Range-FFT? 

[davidtosenovjan]

I am forwarding expert's answer as it is:

"SAF86xx compress Range FFT on BBE32, it is equivalent to SPT PDMA compression. output compressed Range-FFT cube data can be decompressed via SPT PDMA on central Radar MCU, for example S32R45. then do post-processing and generate point cloud.

BBE32 compression format is identical to SPT PDMA compression format, for example CP4D, it keeps eight 7-bit mantissa bits and one 4-bits of exponent for 4 complex data compression, it is a kind of lossy data compression method, more information please check the S32R4x/SAF8xx SPT PDMA chapter, or try PDMA compression example in MATLAB SPT Toolbox. BBE32 instruction extensions can support such compression SW implement, RSDK will most likely provide the compression driver in the future."