Some more details are needed, for example the kind of modulation, and so on.
It seems to be the output of a GMSK demodulator for a system like DECT or Bluetooth with a pseudorandom sequence applied.
if so, the only problem I see is in the average of the curve at the beginning, this can make some mistakes (usually these systems has guard time and known initial preamble and sequences to get rid of it) and maybe the bandwidth of the filter is too low (sequences like 1010 are too low in amplitude, while 11001100 are ok).
But I have gone too far maybe...
I hope it can help.
Mazz