Re: File Sink Block

At https://github.com/jmfriedt/active_radar/ you will find the solution
I have been using for Synthetic Aperture RADAR using GNU Radio. In this
application I wish to collect coherent measurements from both reference
(transmitted pattern) and surveillance (reflected signal) channels when
the antenna location is stable. The GNU Radio part is
https://github.com/jmfriedt/active_radar/blob/master/zeromq_demo_rev1.py
is continuously streaming data using a UDP-like datagram flow based on
0MQ Publish. I selected this approach to make sure any
desynchronization between channels (e.g. using RTL-SDR dongles) remains
constant during the whole measurement and not restarting acquisition
with a random delay for each antenna position.
The antenna position is driven by an external piece of software -- I
like GNU Octave but feel free to use Python -- where event are
triggered in the GNU Radio flowgraph, e.g. moving the antenna, then a
0MQ socket is opened, the processing software fetches the number of
samples needed (here correlation), and then continue to the next
position
https://github.com/jmfriedt/active_radar/blob/master/zeromq_demo_rev1.m

Closing/opening the 0MQ socket is not elegant but is the only way I
found to make sure I fetch the latest data from the new antenna
position and not some leftover data from a previous position.

Best, JM

> I'm forwarding this to the mailing list, so that others can have a
> look at it as well! Let me quickly address a few things:
>
> On 08.08.23 21:32, Henry Powell wrote:
> > The flow graph starts. I collect data and save it to a file with
> > file_sink. I read the file and get the maximum value from the file
> > (differently, I can do this with keep 1 in n, but I would be happy
> > if we proceed through the current situation), then I am saving the
> > maximum value in a list from the data I have saved in the file. I
> > reset the file with the file_sink open and close commands.
>
> Interesting! How do you know when to open / close? I ask because this
> sounds a bit difficult, as, depending on the blocks with which you
> capture and process your data, there might still be samples "in
> flight" after your collection finishes.
>
> > I repeat this maybe 2000 times, varying from
> > measurement to measurement. After recording all the maximum values
> > in a list, the flow graph terminates itself.
> >
> > Writing data to a file and reading from there makes me lose speed.
> > I am looking for an alternative solution to it.
>
> Ah, what about this concept:
>
> ┌───────────┐
> ┌────────────┐ ┌──────────┐ │Make Chunks│ ┌─────────┐
> ┌──────┐ ┌─────────┐ │ │ │ │ │of
> constant│ │Find max-│ │ Head │ │ │ │Data
> Capture│───►│Processing│──►│ "single- │───►│imum per │───►│
> │───►│File Sink│ │ │ │ │ │experiment"│ │
> chunk │ │N=2000│ │ │ └────────────┘ └──────────┘
> │ size │ └─────────┘ └──────┘ └─────────┘ └───────────┘
>
>
> Instead of "taking the data out of GNU Radio" in an asynchronous
> manner, you could look for the maximum within the framework of GNU
> Radio itself. That solves all the file handling overhead.
>
> You can also still save all your experimental data to a file for
> later analysis (just add *another* File Sink, after "Processing" in
> parallel with "Make Chunks…").
>
> > If you're saying that I can do it with a vector sink block, I'm
> > trying it right now. I will let you know the results tomorrow.
>
> Well, the Vector Sink would be easier than the File Sink, but as
> sketched above I think a different solution might be preferable.
>
> > Thank you for your reply. Have a nice day.
> You have good days, as well!
>
> Marcus
>
> > Marcus Müller <mmueller@gnuradio.org
> > <mailto:mmueller@gnuradio.org>>, 8 Ağu 2023 Sal, 11:55 tarihinde
> > şunu yazdı:
> >
> > Hi Henry,
> >
> > that sounds like the description of the Vector Sink block!
> >
> > But in all honesty, maybe you want to take a different approach
> > altogether; really depends
> > on when during the life cycle of a flow graph you need that
> > data. Could you maybe describe
> > how often and when you want to get data from your GNU Radio
> > flowgraph?
> >
> > Best regards,
> > Marcus
> >
> > On 08.08.23 07:26, Henry Powell wrote:
> > > Hello, at the end of my grc I use file sink and then, I read
> > > the data from the file. I repeat this process so many times.
> > > My problem is that writing to files and reading
> > from
> > > them takes too much time, approximately 120ms. I know this
> > > does not seem too much but for my application it's too much.
> > > So, I want to append data to a list and get from
> > there.
> > >
> > > I tried writing an embedded python block via this link:
> > > https://wiki.gnuradio.org/index.php/Embedded_Python_Block
> > <https://wiki.gnuradio.org/index.php/Embedded_Python_Block> but
> > i failed.
> > >  Is there any example embedded python block you wrote about
> > > before? Or a different
> > solution?
> > >
> > > If you can help me, I would really appreciate it.
> >
>



--
JM Friedt, FEMTO-ST Time & Frequency, 26 rue de l'Epitaphe, 25000
Besancon, France