Re: MPSK SVR SNR, Signal and Noise estimation

[Marcus Müller]

Don't think you can directly, since there's no internal notion of S and N kept 
separately
(that's the beauty of this estimator: it's simple).

However, when you have an SNR estimate, you could simply externally compute the 
average
S+N (i.e. the average power of observed samples).

Then, via
S+N = N·(S/N) + N = N·SNR + N = N·(1+SNR)
N = S+N/(1+SNR)
S = S+N - N

you could derive S and N.

Best regards,
Marcus

On 25.10.21 11:54, Moses Browne Mwakyanjala wrote:
> Hi everyone,
> It seems like all estimators in MPSK_SNR_EST class, except for the SVR one 
> (shown below),
> have signal, noise, and SNR estimations.  How does one get the signal and 
> noise
> components from the SVR estimate?
> 
> Regards,
> 
> Moses. 
> 
> mpsk_snr_est_svr::mpsk_snr_est_svr(doublealpha):mpsk_snr_est(alpha)
> 
> {
> 
> d_y1=0;
> 
> d_y2=0;
> 
> }
> 
> 
> intmpsk_snr_est_svr::update(intnoutput_items,constgr_complex*input)
> 
> {
> 
> for(inti=0;i<noutput_items;i++){
> 
> doublex=abs(input[i+1]);
> 
> doublex1=abs(input[i]);
> 
> doubley1=(x*x)*(x1*x1);
> 
> d_y1=d_alpha*y1+d_beta*d_y1;
> 
> 
> doubley2=x*x*x*x;
> 
> d_y2=d_alpha*y2+d_beta*d_y2;
> 
> }
> 
> returnnoutput_items;
> 
> }
> 
> 
> doublempsk_snr_est_svr::snr()
> 
> {
> 
> doublex=d_y1/(d_y2-d_y1);
> 
> return10.0*log10(x-1+sqrt(x*(x-1)));
> 
> }
> 
> 
> Moses Browne Mwakyanjala
> 
> 
> Founder - CEO
> 
> *Remos Space Systems AB*
> 
> m: +46 (0)70 278 2174__
> 
> a: Aurorum 1C,  977 75 __Luleå, Sweden
> 
> w: www.remosspace.com <https://www.remosspace.com/> e: mbkitine@gmail.com
> <mailto:mbkitine@gmail.com> 
> 
> image.png
>