Re: [Discuss-gnuradio] SNR or Other Metrics

[Tom Rondeau]

On Mon, Dec 16, 2013 at 11:22 PM, Dan CaJacob <address@hidden> wrote:
> I'd like to report some signal metrics for satellite downlinks.  SNR would
> seem to be one example, but most algorithms I see for SNR assume that you
> are generating a signal from a separate clean signal and noise source, each
> with known power.  What I have are recordings of real, narrow-band, digital
> FM signals.  I don't have a very strong DSP background.
>
> Can anyone point me to a good textbook that covers such practical
> applications?
>
> Thanks for your help!

Hi Dan,

SNR estimations are quite tricky in real life. If you look at the
digital::mpsk_snr_est blocks, there are multiple implementation inside
that do SNR estimations for M-ary PSK constellations. The different
algorithms have varying accuracy and (mostly) proportional complexity.
But as the name implies, this won't work for FM.

Looking at Google, I found a short paper from Virginia Tech,
co-authored by Steve Ellingson. I know Steve a bit, though he came
just as I was finishing up my work there, and he tends to be pretty
good with these things. So it's likely a good start mathematically for
what you need.

http://www.ece.vt.edu/swe/asmr/docs/ASMR_R16_Kumar2.pdf

Tom