I'm looking for references that explain the trade-offs between more averages with shorter fft sizes or longer fft sizes with fewer averages when generating a spectrogram to perform energy detection of a known signal in AWGN.

As a follow-up to this question, how do you optimize the spectrogram (number of averages, overlap percentage, fft size, window type) for a known signal bandwidth and duration.

My guess is that for tones, a longer fft increases SNR more than averaging but what about other signal types with actual bandwidth?

  • 1
    $\begingroup$ if you have a known signal, why not a matched filter? that would imply a bin center and a DFT length that matches the duration of your signal. windows come into play when you have multiple tones, particularly if there is a tone of interest in the presence of other tones and or colored noise. $\endgroup$ – user28715 Jun 1 '19 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.