RFC: clock synchronization on AFSK modulated signal

[address@hidden]

Thanks to the help I received during the week end, I completed the update
on gr-acars. One thing I realized when working on PSK modulations (e.g RDS)
is the importance of bitrate clock synchronization.

After investigating the M&M algorithm, possibly because of my misunderstanding
of its implementation targeted at identifying zero-crossing of the difference
of the soft bit and estimated hard bit symbol, I have attempted a very naive
clock-rate synchronization on AFSK using maximum likelihood on single bits 
framed by
different bit states (since similar successive bits are hardly affected by 
synchronization issues). I wanted to avoid differentiating the AFSK output
since derivate is prone to noise and might make the zero-crossing identification
unstable (correct ?). I would be quite interested in comments or corrections
to http://jmfriedt.org/acars_clock.pdf or directions to existing literature on
this issue which is far from my field of expertise.

Thanks, JM

--
JM Friedt, FEMTO-ST Time & Frequency/SENSeOR, 26 rue de l'Epitaphe,
25000 Besancon, France