Detailed Analysis

I am looking to do some detailed analysis of my recorded data. I have
been
trying to look at the data in GNURadio using the FFT Plot, but I am
having
a hard time discerning things. Specifically, I am trying to see if a
signal
persists over a period of time. I am going to try and look at things
with
the Waterfall, but I am not sure that will be much better.

I am open to using another tool, such as Octave, if need be.

Any ideas?

Thank you.

I am looking to do some detailed analysis of my recorded data. I have
been

trying to look at the data in GNURadio using the FFT Plot, but I am having
a hard time discerning things. Specifically, I am trying to see if a signal
persists over a period of time. I am going to try and look at things with
the Waterfall, but I am not sure that will be much better.

I am open to using another tool, such as Octave, if need be.

Any ideas?

Hi Paul,

What problems are you facing with the FFT plot? How much data are you
trying to analyze?

If the problem is that the FFT plot exhausts data too quickly, you can
add
a throttle block to slow things down.

Best,
Aditya

Aditya:

I am looking at sections of my data, one-M at a time - that is, from 0
MHz
to 1 MHz, 1 MHz to 2 MHz, etc. So, for example, when I look at 0-to-1
MHz,
I am using an FFT set to FFT-size of 1000. My data is about 13 seconds
long, and I am looking at it at a rate of 1 Hz (so I see about 13
updates).
What I am trying to see is if the data at about nine seconds persists to
the end of the data at 13 seconds. I have tried holding the data at nine
seconds and seeing if things keep matching up, but that is really not
working out for me. I have yet to look at it from a Waterfall
perspective,
which may yield better results. However, in experimenting with the
Waterfall, I am unable to get the “window” to be set the way I can set
it
with the FFT plot. I am getting ready to grab another data set tomorrow,
but I would like to do some analysis prior to getting another data file.

On Wed, Jan 22, 2014 at 7:26 PM, Paul B. Huter
[email protected]wrote:

which may yield better results. However, in experimenting with the
Waterfall, I am unable to get the “window” to be set the way I can set it
with the FFT plot. I am getting ready to grab another data set tomorrow,
but I would like to do some analysis prior to getting another data file.

Hi Paul,

I’m not sure I understand what you’re trying to get to work. Let’s say
you
want to examine the signal over N buckets of bandwidth 1MHz each. Can
you
stream your data into a bank of N band-pass filters in parallel, and
then
examine the FFT output through N FFT plots?

(Sorry, but I don’t think I’m being of much help here.)

Best,
Aditya

Aditya:

Say, for example, that at nine seconds there is a spike at around 15MHz.
I
want to see if that spike stays there until the end of my data file (at
about 13 seconds). I am trying to determine if the signal at 15MHz
sticks
around for awhile. I am setting a “hold” on my FFT at nine seconds,
which
shows up in red, then the FFT keeps going, showing up in blue. I can try
to
“analyze” it by looking at it, but it seems rather inaccurate. I am
trying
to find a better way to do this type of analysis, which may be possible
with the Waterfall, but I think I may need another tool.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Paul,

waterfall seems to be exactly what you want - Fourier transform
amplitude over time.
Although I can see the need for different windows in special
applications, why are you not satisfied with the window choices the QT
as well as the WX waterfall plots offer?
And: Although there is a lot of sense in channel selection prior to
analysis, if you do directly adjacent channels anyway, and try to
correlate what happens in different channels, why not keep the full
bandwidth and just increase the DFT size instead? Since temporal
resolution constraints seem to be in the order of 0.1s, at your
physical sampling rates (N*1Msam/s) that offers large (>>1024) sizes.
Don’t worry: Your computer should not take very long to calculate
20000-FFTs :slight_smile: and you’re doing this offline, anyway.

Anyway I think you’re right, having a dynamically generated waterfall
is most probably not the right tool.
My weapon of choice for the analysis of offline (recorded) data is
python:
Use the numeric data types of numpy[1], the algorithms in scipy, the
plotting functionality of matplotlib.pyplot [2], and the python shell
of my choice is bpython, although Ipython usually integrates better
with matplotlib[3].

Happy understanding your signal,
Marcus M.

[1] the same that GR uses
[2] strikingly similar to matlab, but not smothered by a terrible
programming language
[3] If you’re on the lookout for a little Matlabeske GUI, try
enthought canopy.
On 23.01.2014 01:26, Paul B. Huter wrote:

me. I have yet to look at it from a Waterfall perspective, which

I am open to using another tool, such as Octave, if need be.
you can add a throttle block to slow things down.

Best, Aditya

_______________________________________________ Discuss-gnuradio
mailing list [email protected]
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJS4MRuAAoJEAFxB7BbsDrLXC8H/11+IPSyMqnRBwNT0FmOINU0
IshtLj6pUfp+/guv5mevNk5GuxpOggp1uAr8r0tkglDAcM4u1jrVYXNt40lSwGbH
ESXAukZAJOAlVq9Kd69DdsEwag7GE2NpBaNQPXrREdJ3SqpgwIlY14V+g3Rp/s3q
xKh2KyPz9jW5SFEQnLcaaohcYNlqyBTQj0cDGcg+160NwLoFgYydB4I3V+tVJVWt
gpRqng4keb5/wvD7SKXRB9xbco3q7+7eeAs1B+6BAe7pt2/wh/244np/3KAYk918
vy4cMmywUoiKFzCwytwQDTGSSEWjKWZ7MQYHoRNyo8UiTVMuS62anQ5eti5HrtA=
=ZvH4
-----END PGP SIGNATURE-----

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Forgot to mention: Try the gr_plot_fft[…] tools that come with
gnuradio, they basically do the python stuff for you. Look at the
options to set physical sampling freq etc.

On 23.01.2014 08:27, Marcus M. wrote:

temporal resolution constraints seem to be in the order of 0.1s, at
better with matplotlib[3].

from 0 MHz to 1 MHz, 1 MHz to 2 MHz, etc. So, for example, when
ready to grab another data set tomorrow, but I would like to do

I have been

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJS4MV1AAoJEAFxB7BbsDrL7L8H/37zHS87esDN1E7syQjg8odB
X0oweyC6IJ6UFTBlp63HLpbvNyehlL1CkwNk3/La38ynJXNzJbgnVpvgzLOwZApj
oUz2XZxdEPkKJTtSUdU8PvYtr4sVY9VJ+kr2NvOarhVOXlK3k9YvkELvz35tBvxC
l9Yw2DzFluLOuTHiyVOW0N+4+SoFx4Y+pG8TkGymjr7l2Kfeh2chgPGm0l9ZD1wL
Ke7/EC2rwerF37mT+Q8jGSVU6KVwC6oKH+fcppgeY0u44MlUQszQZauPFTE40kSJ
T4QMrbbPmSODLVYI76cSZUzj4VpbyYgDpUevHCcYwBL5nirO+2IM3NN0jyxkpaw=
=yUdb
-----END PGP SIGNATURE-----

This forum is not affiliated to the Ruby language, Ruby on Rails framework, nor any Ruby applications discussed here.

| Privacy Policy | Terms of Service | Remote Ruby Jobs