Re: [wsjt-devel] A technical question..

[Joe T. <jo...@pr...>]

Hi David,

Welcome to this list!

On 8/25/2017 9:20 AM, David Alloza wrote:

> I would like to know how you evaluate the noise level used for the SNR
> calculation, it seems to me that there are two ways to do it:
> 
> 1> You take the average spectral power present in the spectrum received (on
> 2.4Khz bandwidth) , and you compare to the power of the station received.
> 
> It's simple, but the presence of several powerful stations will change the
> result.
> 
> 2> or you are trying to evaluate the background noise level in the spectrum
> of 2.4Khz by hide the influence of the stations received, and in this case
> how is this evaluation done?

SNR calculations are done in somewhat different ways for each of the 
WSJT-X modes, but all are intended to be a close approximation to your 
#2.  Briefly stated, we fit a "baseline" to the spectral regions in the 
Rx passband that do not contain signals.  The spectral density of noise 
close to a signal's frequency is evaluated from this baseline.  This 
value is scaled up to 2500 Hz bandwidth, and becomes the denominator in 
the S/N calculation.

For interest: at present, calculation of SNR in the new FT8 mode comes 
close to the above ideal only for signals *in the clear*.  SNRs for 
signals with nearby interference are effectively evaluated as

   Signal/(noise + interference)

We'll do a better job in an upcoming release.

	-- 73, Joe, K1JT.