Hi there, I am quite newbie in the field of signal processing, thus, this question may look silly but: - Is there a way to estimate SN ratio of signal with a priori unknown parameters? Let's say, I have received an AM/FM signal and need to estimate SNR of it. I have no any klue about characteristics of transceived signal nor communication channel. Is it feasible? Thank you in advance, Igor.
SNR of a signal with unknown parameters
Started by ●October 29, 2008
Reply by ●October 29, 20082008-10-29
On Oct 29, 9:00�am, "ivli" <i...@pochta.ru> wrote:> Hi there, > > I am quite newbie in the field of signal processing, thus, this question > may look silly but: > - Is there a way to estimate SN ratio of signal with a priori unknown > parameters? � > Let's say, I have received an AM/FM signal and need to estimate SNR of it. > I have no any klue about characteristics of transceived signal nor > communication channel. Is it feasible? � > > Thank you in advance, > Igor.That's not really possible. Knowing that there's an AM transmitter, at some unknown place in the universe, with unknown transmit power, antenna pattern, receiver characteristics, etc. doesn't give you any information on what the SNR would be if you received it. You need to know *something* in order to start making conclusions. Jason
Reply by ●October 29, 20082008-10-29
On Oct 29, 6:00�pm, "ivli" <i...@pochta.ru> wrote:> Hi there, > > I am quite newbie in the field of signal processing, thus, this question > may look silly but: > - Is there a way to estimate SN ratio of signal with a priori unknown > parameters? � > Let's say, I have received an AM/FM signal and need to estimate SNR of it. > I have no any klue about characteristics of transceived signal nor > communication channel. Is it feasible? � > > Thank you in advance, > Igor.possible with autocorrelation..plz read into depth of autocorrelation literature..
Reply by ●October 29, 20082008-10-29
On Oct 29, 8:00 am, "ivli" <i...@pochta.ru> wrote:> Hi there, > > I am quite newbie in the field of signal processing, thus, this question > may look silly but: > - Is there a way to estimate SN ratio of signal with a priori unknown > parameters? > Let's say, I have received an AM/FM signal and need to estimate SNR of it. > I have no any klue about characteristics of transceived signal nor > communication channel. Is it feasible? > > Thank you in advance, > Igor.Why are you asking this question? Are you trying to build something?
Reply by ●October 29, 20082008-10-29
"ivli" <ivli@pochta.ru> writes:> Hi there, > > I am quite newbie in the field of signal processing, thus, this question > may look silly but: > - Is there a way to estimate SN ratio of signal with a priori unknown > parameters? > Let's say, I have received an AM/FM signal and need to estimate SNR of it. > I have no any klue about characteristics of transceived signal nor > communication channel. Is it feasible?No. You can't tell the difference between, e.g., a DSSC transmitter sending noise over a noiseless channel and a DSSC transmitter sending nothing over a noisy channel. -- % Randy Yates % "Bird, on the wing, %% Fuquay-Varina, NC % goes floating by %%% 919-577-9882 % but there's a teardrop in his eye..." %%%% <yates@ieee.org> % 'One Summer Dream', *Face The Music*, ELO http://www.digitalsignallabs.com
Reply by ●October 29, 20082008-10-29
Thank you for your answers, I had some gnawing doubts despite engineering background told me it's impossible. Regards, Igor.>"ivli" <ivli@pochta.ru> writes: > >> Hi there, >> >> I am quite newbie in the field of signal processing, thus, thisquestion>> may look silly but: >> - Is there a way to estimate SN ratio of signal with a priori unknown >> parameters? >> Let's say, I have received an AM/FM signal and need to estimate SNR ofit.>> I have no any klue about characteristics of transceived signal nor >> communication channel. Is it feasible? > >No. You can't tell the difference between, e.g., a DSSC transmitter >sending noise over a noiseless channel and a DSSC transmitter sending >nothing over a noisy channel. >-- >% Randy Yates % "Bird, on the wing, >%% Fuquay-Varina, NC % goes floating by >%%% 919-577-9882 % but there's a teardrop in hiseye...">%%%% <yates@ieee.org> % 'One Summer Dream', *Face The Music*,ELO>http://www.digitalsignallabs.com >
Reply by ●October 29, 20082008-10-29
On Wed, 29 Oct 2008 08:00:17 -0500, "ivli" <ivli@pochta.ru> wrote:>Hi there, > >I am quite newbie in the field of signal processing, thus, this question >may look silly but: >- Is there a way to estimate SN ratio of signal with a priori unknown >parameters? >Let's say, I have received an AM/FM signal and need to estimate SNR of it. >I have no any klue about characteristics of transceived signal nor >communication channel. Is it feasible? > >Thank you in advance, >Igor. >For the most part, yes, this can be done, if the signal is reasonably well isolated in the spectrum that it occupies. The easy way is to look at the signal in the frequency domain. The distance from the signal peak power to the noise floor is (S+N)/N. It's straightforward to compute S/N from that. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Reply by ●October 29, 20082008-10-29
On Wed, 29 Oct 2008 10:21:03 -0700, Eric Jacobsen <eric.jacobsen@ieee.org> wrote:>On Wed, 29 Oct 2008 08:00:17 -0500, "ivli" <ivli@pochta.ru> wrote: > >>Hi there, >> >>I am quite newbie in the field of signal processing, thus, this question >>may look silly but: >>- Is there a way to estimate SN ratio of signal with a priori unknown >>parameters? >>Let's say, I have received an AM/FM signal and need to estimate SNR of it. >>I have no any klue about characteristics of transceived signal nor >>communication channel. Is it feasible? >> >>Thank you in advance, >>Igor. >> > >For the most part, yes, this can be done, if the signal is reasonably >well isolated in the spectrum that it occupies. > >The easy way is to look at the signal in the frequency domain. The >distance from the signal peak power to the noise floor is (S+N)/N. >It's straightforward to compute S/N from that. > >Eric Jacobsen >Minister of Algorithms >Abineau Communications >http://www.ericjacobsen.org > >Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.phpSmall clarification: there's some discretion in what I mean by "peak power" above. If the signal spectrum is well-behaved with a nice, flat top, that's easy to get right. If it's NRZ or something with a non-flat top to the signal spectrum, then use some discretion in picking the "peak" level or a reasonable representation of the signal power. This also applies to signals with big pilot tones sticking through or something like that. Eric Jacobsen Minister of Algorithms Abineau Communications http://www.ericjacobsen.org Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php
Reply by ●October 30, 20082008-10-30
Hello Eric, Thanks, I guess it should help since I need only rough estimate of SNR value. Regards, Igor.>On Wed, 29 Oct 2008 10:21:03 -0700, Eric Jacobsen ><eric.jacobsen@ieee.org> wrote: > >>On Wed, 29 Oct 2008 08:00:17 -0500, "ivli" <ivli@pochta.ru> wrote: >> >>>Hi there, >>> >>>I am quite newbie in the field of signal processing, thus, thisquestion>>>may look silly but: >>>- Is there a way to estimate SN ratio of signal with a priori unknown >>>parameters? >>>Let's say, I have received an AM/FM signal and need to estimate SNR ofit.>>>I have no any klue about characteristics of transceived signal nor >>>communication channel. Is it feasible? >>> >>>Thank you in advance, >>>Igor. >>> >> >>For the most part, yes, this can be done, if the signal is reasonably >>well isolated in the spectrum that it occupies. >> >>The easy way is to look at the signal in the frequency domain. The >>distance from the signal peak power to the noise floor is (S+N)/N. >>It's straightforward to compute S/N from that. >> >>Eric Jacobsen >>Minister of Algorithms >>Abineau Communications >>http://www.ericjacobsen.org >> >>Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php > >Small clarification: there's some discretion in what I mean by "peak >power" above. If the signal spectrum is well-behaved with a nice, >flat top, that's easy to get right. If it's NRZ or something with a >non-flat top to the signal spectrum, then use some discretion in >picking the "peak" level or a reasonable representation of the signal >power. This also applies to signals with big pilot tones sticking >through or something like that. > > >Eric Jacobsen >Minister of Algorithms >Abineau Communications >http://www.ericjacobsen.org > >Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php >
Reply by ●October 30, 20082008-10-30
On Oct 30, 3:58 am, "ivli" <i...@pochta.ru> wrote:> Thanks, I guess it should help since I need only rough estimate of SNR > value.What it comes down to is your ability to determine by eye or by algorithm what constitutes signal and what constitutes noise. Once you can do that, the ratio is relatively easy. But since you stated the problem as being essentially an arbitrary unknown signal, people are going to point out that it's impossible to determine which is which. For old modulation schemes it isn't too hard, but modern efficient modulations look increasingly like noise (conceptually speaking, ideally encrypted data transmitted with ideally efficient modulation would be indistinguishable from noise to anyone not in on the secret of how to demodulate and decrypt it)
