Re: [Discuss-gnuradio] Integration time

[Johnathan Corgan]

On Mon, Dec 15, 2008 at 9:47 AM,  <address@hidden> wrote:

> Fairly new to GNU Radio.  Have a USRP with DBSRX connected to a Linux box.  
> What
> my professor and I are trying to figure out is what the data being collected
> represents in terms of the integration parameters.  I probably don't know how
> to formulate the question correctly, but we're able to set integration time,
> size of FFT, and FFT Rate.  What we really want to know is how to relate all
> these things into real time- if I set these parameters what am I really 
> getting
> in terms of integration time in seconds, if we look at one minute's worth of
> data, with a certain number of channels, what does the data really represent 
> in
> terms of the parameters, etc.  If anyone has any suggestions or documentation 
> to
> point me to, I'd appreciate it.  Thanks.

There really isn't any true integration going on in usrp_fft.py, if
that is what you are referring to.

The GNU Radio flowgraph portion of this application slices the input
data stream from the USRP into vectors of 1024 samples in length.
Then, it only keeps "one in n" of these vectors such that there are
only ~15 (by default) new vectors per second.  Then, these vectors go
through a windowed FFT and are displayed in the main window.

When averaging is turned on, there is an exponential averaging filter
applied to each bin.  By default the tap value is 0.1, though it is
adjustable in the GUI.  So the term "integration time" concept doesn't
really apply.

-Johnathan