Rik Strobbe wrote:

[snip]
If the FFT window slides over a transmitted dot the FFT result (for that
specific frequency) will increase, reach a maximum when the FTT bin
completely overlaps with the dot and then will decrease again. With other
words, a certain pattern will be created and if we know when the dot begins
(synchronisation) and the duration of the dot it might be possible to do a
kind of pattern recognition.
[snip]

Rik,
      IMHO, if the SNR is high enough to allow for such a pattern recognition,
then you won't have any need for it...  if the goal is to be able to extract 
dots and
dashes from static crashes, it is hard to beat the pattern recognition features 
of
the humain brain. But I may be plainly wrong (it happened before...)

73
Alberto   I2PHD