> 27 hours will not be easy.

Indeed, four terminators to cope with.  Half the message
will have daytime phase and half night time phase and
a mess in between.  No chance of a decode if day and night
phase differs by more than 90 deg (which it probably will).

So I will have to get a nominal phase from LWPC or my
own modeling program (or just guess) and use that to EQ the
signal before decode.

An interesting challenge!  Thanks Stefan

> Start: 29.04.2016  16:00:02
> Period: 40
> Length: 45 (!)

No problem with the length, and for future tests like
this, why not run a 16K code (with half symbol period)
and with longer constraint length, eg 16K25A?  The
decoder will need about 100 Gbyte RAM for that, but
that's ok, I just use a big AWS Xeon with 36 cores and
a ton of RAM.   The extra bit of coding gain is worth it
and might make all the difference.

--
Paul Nicholson
--