The following call is how I read data into MATLAB from the GNURadio file
sink. I think this would almost work in Octave as well but I haven't tried.
If not it, it at least illustrates the proper format.
FileID = fopen('Filename','r');
import_channels = 2;
SamplesPerFrame = 1024;
import_format = 'float32';
x = fread(FileID,[import_channels SamplesPerFrame],import_format)';
complex_baseband = complex(x(:,1),x(:,2));
I have this embedded in a MATLAB system class. I extricated the
fopen/fread() calls for clarity.
HTH
--
Martin K.
On Thu, Jun 7, 2018 at 9:30 AM, Federico 'Larroca' La Rocca <
flarr...@gmail.com> wrote:
> Hello again,
>
> To export from GNU Radio to octave, you should first use a "File Sink"
> block and then one of the scripts in gr-utils/octave. I'm pretty sure they
> are compatible with matlab too.
>
> About the SNR, that depends on what you want. If you want the power of
> noise that actually affects your decision system divided by the power of
> the symbols, then MER is what you are looking for (assuming your sampling
> and frequency correction works properly). If you want the more classic SNR
> (i.e. power of the useful signal divided by the total noise power in the
> band), then that's more involved. You may try the following: since you know
> the constellation points and the pulse, you may easily calculate the power
> of the useful signal (assuming your AGC works properly and that the channel
> is flat fading). The difference between this power and the actual received
> power is due to noise (assuming independence).
>
> However, what matters performance-wise is the amount of noise that gets
> into the decision system, and not the actual SNR.
>
> Hope it helps!
>
>
>
_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
