GMSK midamble template matching

GNU Radio
1

I would like to identify the 26 midamble bits in each burst of GSM
(GMSK)
data. I hear that a lot of ppl use correlation to do this. However, I
am
struggling to understand how this technique could work due to the ISI
from
the Gaussian pulse shaping. It seems that the wavefrom of the 26
midamble
bits is depdented on the data coming before them and thus, template
matching
would not work. There must be something I am missing here. Any help
would
be greatly appreciated.

Thanks in advanced,
Isaac

View this message in context:
http://www.nabble.com/GMSK-midamble-template-matching-tp17512969p17512969.html
Sent from the GnuRadio mailing list archive at Nabble.com.

2

Isaac,

You are right, the midamble waveform is dependent on the data bits
occuring
before the midamble. However, the dependency is just a constant phase
offset. Therefore, correlation with the GMSK modulated midamble bits
will
still result in good correlation scores. One more point is that the bit
just before the midamble can cause some uncertainty in the waveform at
the
very beginning. You can resolve this by correlating with only the
middle 16
bits of the midamble instead of using the entire 26.

Hope this helps,
Ben

3

I am also finding that I need to match against not only the template,
but the
complex conjugate of the template too!

–
View this message in context:
http://www.nabble.com/GMSK-midamble-template-matching-tp17512969p17541609.html
Sent from the GnuRadio mailing list archive at Nabble.com.