[Discuss-gnuradio] RE: [Commit-gnuradio] gnuradio-core/src/lib/general gr_correlate_acce...

[Tom Rondeau]

[Moving to list, per Eric's request]

Right, so we are talking about implementing 8PSK and pi/4 DQPSK, the later
being more important as it shows up in implementation often. My concern with
both, though, is the phase and frequency synchronization. I'm sure someone
out there has some experience with this.

To summarize, we have a second order Costas loop for BPSK and a fourth order
Costas loop for QPSK. What is normally used for synchronization with these
modulations? For pi/4 DQPSK, I was thinking we could use the current fourth
order loop, in which we can set the desired rotation. Each symbol received
would cause the rotation variable to flip 45 degrees.

a) does this work, and
b) is there a better way to do this?

Tom



> -----Original Message-----
> From: Johnathan Corgan [mailto:address@hidden
> Sent: Wednesday, June 28, 2006 3:21 PM
> To: Tom Rondeau
> Cc: Eric Blossom
> Subject: Re: [Commit-gnuradio] gnuradio-core/src/lib/general
> gr_correlate_acce...
> 
> Tom Rondeau wrote:
> 
> > My next target modulation is QPSK, OQPSK, and pi/4 DQPSK. If you give me
> > information on some implementation details for 8PSK, I'll work those in,
> > too.
> 
> pi/4 DQPSK *is* 8PSK, but with a restriction.
> 
> 
>     1
>  8     2
> 
> 7        3
> 
>  6     4
>     5
> 
> The only allowed transitions are from an even constellation point to an
> odd one, or vice versa.  For example, from point 1 it goes to 2,4,6,8
> based on the two bits being encoded.  There is only ever a phase change
> of pi/4 or 3*pi/4 (hence the name).  This prevents the RF envelope from
> dropping to zero (as would be the case of a transition from say, 1 to 5,
> or from 7 to 3).
> 
> You're right, the carrier/phase tracking is the hard part.  Since it's
> differential you can let the lock be on any of the eight orientations;
> the symbol is encoded in how much phase change there is from one
> successive symbol to another.
> 
> (Eric, I didn't copy you on my first email to Tom, asking about pi/4
> QPSK.  And this isn't copied to the list as I don't know how much you
> guys are sharing at this point.)
> 
> -Johnathan