0
$\begingroup$

What are the standard ways to measure SNR, particularly before demodulation/synchronization?

$\endgroup$
1
$\begingroup$

Perform a 2 dimensional cross correlation against a noise free replica of the signal you expect to receive with delay offset on one axis and frequency offset on the other, and find your maximum correlation. Normalize that to get $\rho$ and from $\rho$ you can get SNR. Do this after channel equalization, otherwise you will get the SNR degradation due to ISI as well.

For further details, see:

Noise detection

and

How can I find SNR, PEAQ, and ODG values by comparing two audios?

If the SNR is high and the noise is well behaved (thermal noise floor, no band selective interference present) then you can make a reasonable estimate directly from a spectrum plot such as the example I show below, where the SNR in this case was 40 dB as is clear from the distance between the signal and the noise. This is signal power to the total noise power that is in the signal bandwidth.

PSD

$\endgroup$
  • $\begingroup$ Yes that makes sense, but what if you don't know the signal you will receive? Only its approximate/average bandwidth? $\endgroup$ – FourierFlux Dec 15 '19 at 2:22
  • $\begingroup$ Can you limit your question to only being high SNR's? $\endgroup$ – Dan Boschen Dec 15 '19 at 2:24
  • $\begingroup$ Maybe, I don't have an explicit application but just a general curiosity. For example: in high SNR applications you can have your adaptive synchronization routines increase their convergence rate. In low SNR applications you want your adaptive elements to change slowly(be more noise immune). So the question is how do you detect SNR without having synchronized the system beforehand? $\endgroup$ – FourierFlux Dec 15 '19 at 20:17
  • 1
    $\begingroup$ Interesting — if you don’t know anything about the signal I not really sure how you can distinguish it from noise—- but usually you know something about the signal that you can exploit at which point it is a matter of being clever about that (usually some sort of correlation is involved) $\endgroup$ – Dan Boschen Dec 15 '19 at 20:44

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.