[Discuss-gnuradio] Bit Error Rate Testing?

[Steven Clark]

I have been experimenting with the gmsk demodulation block, and am able to successfully demod 50M+ continuous pseudorandom symbols at 2.0 MSym/sec. Kudos to the USRP + GNURadio devs for making that possible, and thanks for making GNU Radio open-source.

The testing methodology so far has been to output the demodulator to a file sink, eg:

self.gmsk_demod = gmsk_demod(self, samples_per_symbol=2, verbose=True, log=False)
limiter = gr.head(gr.sizeof_char, 10000000)
dst = gr.file_sink(1, 'output.dat')
self.connect(self.u, self.gmsk_demod, limiter, dst)

then to read the file into matlab and cross-correlate against the known bit sequence.

However, the time has come to start doing some bit error rate tests at various SNRs, and the previously mentioned methodology starts to become impractical for low BERs (given that you generally want to wait until you get 100-1000 or so errors).


SO: is there any way to get the demodulator's output out to a pin without seriously harming the current performance of the demodulator? I have a 2nd USRP available 
if that opens up additional options.

Thanks for any advice.

-Steven