discuss-gnuradio
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] embedding


From: Marcus Müller
Subject: Re: [Discuss-gnuradio] embedding
Date: Fri, 04 Apr 2014 11:47:29 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.3.0

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

Hi George,

I think you're talking about Matlab, aren't you?
Matlab has a resource hungry interpreter for its scripting language.
Embedding that will be possible.
Matlab *can* (in many cases) generate C/C++ code, but that will also
only work with a fully-fledged OS underneath; also, the
Matlab-supplied compiler links against a matlab supplied standard
runtime, which effectively eliminates portability.

Generally, comparing Matlab to GNU Radio is a lot like comparing
apples and pears -- you can do signal processing with both, but the
idea behind is quite different.

GNU Radio is a C++/python-based framework. As such, it needs a runtime
to work on; you won't get far without an OS (memory
allocation/management, threading etc are all in especially heavily used)
You can actually generate GNU Radio applications without using python,
thus reducing the need for dynamicly loaded libraries heavily.

So to be realistic: Embedded OS for GNU Radio boils down to using
Linux, for example in the shape of OpenEmbedded/Yocto. That has been
proven numerous times to work quite well.

I haven't actually tried, but in theory you should be able to
statically link with GNU Radio. I honestly don't know if you win much
by that -- I imagine a embedded platform that's powerful enough to do
useful RF signal processing will most probably not be limited by a few
MBs of superfluous libraries lying around; but maybe I'm wrong about that.

If you want to see an embedded GNU Radio based Signal Processing
hardware in action, there are several people that use raspberry Pi's
together with RTL-SDR dongles, or take a look at the Ettus Embedded
USRP series -- these are ARM platforms running an Angstrom Linux
distribution.

So to answer to your considerations:
The GNU Radio approach is usually to develop your application on your
workstation, and run it on an embedded device that's powerful enough.
It's been proven to work well :)

Greetings
Marcus

On 04.04.2014 10:20, George Refseth wrote:
> Hi
> 
> Anyone with input as to the feasibility of maintaining either
> gnu-radio or math-lab design and have something embeddable without
> pulling the whole environment into the target? Generating c or c++
> to be compiled for the target with or without a real-time os.
> 
> What I foresee as a problem is the transfer of design from
> modelling to run-time, and that becomes a one-way road and only
> done once so that one looses the value of the model be it gnu-radio
> or math-lab based.
> 
> best regards, George Refseth
> 
> _______________________________________________ Discuss-gnuradio
> mailing list address@hidden 
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQEcBAEBAgAGBQJTPn+xAAoJEBQ6EdjyzlHtU1oH/jQaYIfIvpa6q1OeXOWbRdf8
xrYWAlpyfZTd7sg4sX081n2abfhH67qV5fT2OnuD8QGjPygwI6/AI/ptLGPNp+6C
EnUS5HlEv5q1PHh1HZiKfPouUZv9aq1BZrtFi2G19OXpGyK8rV65anyahta7B4lw
AlhPTgjtbsEG+yeub0zrhjQU+rZjjvYP3q+rFuiyYBGuJGeN+tLMAFZdlZXk0q6V
4wRvZGM4C9goCDwuhpdI+C7/N9ZX/Tys+0FlagSKcCHn9b8Ns5+VfemlzDIzUo/s
1Qk4OlTo+78bJ/RqspnbP5AtwVOy9cSA4B7yJHcur10OeyYEukMAnNBeOf3MxM4=
=A7NH
-----END PGP SIGNATURE-----



reply via email to

[Prev in Thread] Current Thread [Next in Thread]