I could finally find the correct configuration. I just added a "pack k bits" block between the "gmsk demod" and "correlate access code" blocks. I found it while I was monitoring and comparing the input and output waveforms of the mod and demod blocks. As was written in the gmsk demod document, this block produces a stream of bits, one bit per byte and the bit is written on the LSB of the corresponding byte. So I searched for a block that extracts one bit from a byte and the result was "pack k bits" which takes one bit from every input byte and puts the k bits that have been taken, in a row beside each other. It worked!
But when I saw the spectrum of the modulated signal, i.e the output of the gmsk mod, it was not a well-known bell-shaped spectrum around zero Hz! It was like the picture below. Why is it like that?
Sorry for too many questions, can I also ask how to set the right sampling rate of the USRP sink? I mean according to a flow graph like the below transmitter, how can I calculate the sufficient sampling rate of the USRP sink? Any hint is appreciated.
Thank you very much.
--
Amir Hossein Aalipour
MSc student of Electronics engineering
Department of Electrical Engineering
Sharif University of Technology
modified_loopback.png
GMSK_spectrum.png
Final_transmitter.png