• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 2146 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AWR1642BOOST: Zoom FFT Algorithm

Avinash Kondeppady
Prodigy 40 points
Part Number: AWR1642BOOST
Other Parts Discussed in Thread: AWR1642

Hi,

I was going through the zoom fft experiment for the awr1642.

On going through the user guide it says that the algorithm is based on the general factorisation of a FFT of size N  into N1 number of N2 size ffts.

However most of the literature surrounding zoom fft is based on a band pass filter followed by a decimation. 

Can someone please explain further on the algorithm used by TI and if it is different/similar to the algorithm mentioned above?

Regards

Avinash

  • 0
    TI__Genius 11405 points

    Hi Avinash,

    Zoom FFT in TI mmwave sensor lab takes two steps. First is to run a coarse 1D fft and search for the peak. The 2nd step is to zoom in only around the peak detected in step one.

    The twiddle of fine (zoom-in) fft is generated (exp(-j2pik/N) where N = N1xN2. Multiply a set of twiddles to the input signal and find the indices. The related code implementation is in RADARDEMO_highAccuRangeProc_priv.c. 

    Let me know if you have any questions.

    Thanks and Regards,

    Michelle

  • 0 in reply to Michelle Liu
    Prodigy 40 points
    Is there any specific reason for which TI has gone for this algorithm as opposed to others. Is it faster to implement/better results compared to other algorithms?
  • 0 in reply to Avinash Kondeppady
    TI__Genius 11405 points
    Hi Avinash,

    The filter design cost more time for implementation. Also cycle cost (usually require multiple stages) on DSP will be bigger for filter method than FFT zoom-in as we did. It in turn costs more power. In the end, major concern is the design cost. If you have a good filter design, that can be implemented.

    Regards,
    Michelle