Re: [Discuss-gnuradio] Analyzing FFT for multiple maxima

[John Clark]

Paul Mathews schrieb:
> Here's a screen shot of FFT output:
>
> http://i394.photobucket.com/albums/pp28/optoeng/2peaks.jpg
>
> The image is typical, in that there are multiple narrow local maxima, but
> the amplitude of these particular peaks is quite a bit higher than I expect
> 'in the field'. I'm trying to work out an efficient way to compute the
> locations and relative amplitudes of these maxima. Can someone please
> suggest either an algorithm and/or the most similar gnuradio application
> that I can examine as candidates for this analyis? Thanks.
> Paul Mathews

I thought Analog TV was going of the air in the US... The difficulty is 
that when the FFT is taken
from the real world, and the signal tends to the noise level, most 
'peak' detection algorithms fail,
well, have lots of noise in the 'accept' condition...

So, long term averaging, and 'knowing' that a peak would appear at 
offset x from the right or
left side of the spectrum will be more likely to succeed.

John Clark.