Re: [Discuss-gnuradio] SDR-Based BERT

[Tom Rondeau]

On Tue, Jun 25, 2013 at 11:50 PM, Dan CaJacob <address@hidden> wrote:
> I often have to test hardware RXs and though I have a hardware BER test
> fixture, it's kind of ancient and difficult to work with.  I have fantasized
> about using a USRP to transmit a PRB (pseudo-random bit-sequence) to a RX
> modem.  The data would then be brought back into GnuRadio and compared with
> the originally transmitted PRB to determine the actual BER.
>
> To align the data, I figured I could correlate the sent code against the
> received data to determine any offset and align the data for comparison.
> The correlation should be pretty forgiving to bit errors if the right PRB is
> chosen, I think.
>
> My question is: does anyone have additional thoughts about this?  I don't
> think that this sort of tool exists yet in GnuRadio.  I have seen BER tools
> and even a nice BER block from Balint, but I think they all assume the data
> is already aligned.  I imagine that Balint's block could still be used if
> the alignment was done first.  Am I missing any other tools out there?
>
> As for getting the received data back into the flowgraph, I imagine that
> this might require a new source, like a synchronous serial port.  I have
> used libmpsse to work with some of the MPSSE-enabled FTDI chips before.
> Maybe that could be implemented as a source block.  My other thought was to
> use a LFTX board and bring in the clock and data as separate RX channels
> then proceed from there.
>
> Thanks!

Hey Dan,

That's a good problem. My only recommendation at this point is the
correlate_and_sync block that I'm working on for burst modems. Started
it at our last hackfest. You can find the code on my github repo
(https://github.com/trondeau/gnuradio.git) in the 'grhack13' branch.
It's working in a very specific case but is close to working
generically. It sounds close to what you want.

Tom