Re: [Discuss-gnuradio] problem with packet decoder for gmsk

[John Malsbury]

To expand on Martin's comments, here are some of the issues that *could* be present:

• Add a mult_const block or otherwise reduce the amplitude of the i/q signal going to the usrp_sink from +/- 1.0 to 0.5 or 0.7.  This ensures you are not non-linear in the transmitter.
• tx and rx_gain -> if you have these two devices a few feet apart (3-5 dB for both tx and rx should work)
  
• Frequency errors between the two USRP devices (or is this on a single device?)   Ultimately, you could include some automatic, course frequency correction
  
• In my experience, the GMSK demod will lock within 100-200 symbols.  Since this is a continuous streaming flow graph driven by the random source, transient behaviour shouldn't be an issue.  You should have a continuous stream of samples to lock to.  So I don't think lock/sync time for the clock recovery is an issue.
  
• Have you changed any parameters on the packet encoder/decoder blocks from the defaults?  What is the bit/symbol and samps/symb set to?  What's the samps/symbol set to on the gmsk mod and demod?
  
• Instead of a random source, test with a vector source and a known pattern.
• To give a quick sanity check on your encoder/decoder config, connect the two *without* any channel.  If that doesn't work, you definitely have some config issues.

Also, once you're through with this experiment, you might consider using gr-mac, which provides similar functionality.  Balint and I have been hacking on it late tonight, you can find the latest here:

https://github.com/balint256/gr-mac

Also, a known working version that is a bit older and less bleeding edge:

https://github.com/jmalsbury/gr-mac

http://conferences.sigcomm.org/sigcomm/2013/papers/srif/p31.pdf

-John

On Thu, Mar 27, 2014 at 1:52 AM, Martin Braun <address@hidden> wrote:

On 03/26/2014 12:30 PM, Ank wrote:
> Hello all,
>    I'm working on gmsk transmitter and receiver past 6 months , the problem
> that I'm facing is in the packet decoder.
>
>   I tried with wav file and signal source it worked good.
>
>   my block is of this type
>
>
> Random_source=>Packet_encoder=>gmsk_modulator=>rational_resamlper_1=>uhd_sink
>
>
> uhd_source=>rational_resampler_2=>low_pass_filter=>gmsk_demodulator=>packet_decoder=>scope_sink/file_sink

There are a lot of things that can go wrong in this setup. Maybe the
sync is not locking fast enough, and destroying the preamble. It's hard
for us to figure this out from a distance.

Try replacing the UHDs with a channel model, and go crazy with that. See
what happens.

M


_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio