Re: [Discuss-gnuradio] OFDM demodulation problem / example video

[Jens Elsner]

Further more this means that the demodulated symbol (differences!) should
always have phase 0, pi/4, 2pi/4, 3pi/4 (plus a *constant* offset of pi/8).
But this offset is varying.

Am I right?

And why should the transmitter introduce a varying offset? The problem is 
somewhere in the demodulation chain.

Jens

On Mon, Apr 03, 2006 at 11:23:49AM -0400, Robert McGwier wrote:
> Jens:
> 
> Then this means that the demodulated symbol should always be on points  
> abs(X) = abs(Y) = 1.  HOWEVER,  this must be built correctly into the 
> transmit symbols BEFORE they are transmitted for them to be received by 
> the simple division you have done below.  I hope I get time to visit it 
> in greater detail today.
> 
> Bob
> 
> 
> 
> Jens Elsner wrote:
> >Bob, thanks for the reply.
> >
> >This is exactly what I am doing:
> >
> >[...]
> >    dem = data ./ data_l;
> >[...]
> >
> >A point-wise division by the last OFDM symbol. This also restores
> >amplitude, but has the same effect on the phase.
> >
> >I do not control the receiver, my data is from a radio station. I am
> >only trying to demodulate. I am synchronised in the 
> >frequency domain. Time domain synch is also appropriate.
> >
> >I guess I'll just have to get new data to verify this. It seems just too
> >odd.
> >
> >Jens
> >
> >On Mon, Apr 03, 2006 at 10:48:32AM -0400, Robert McGwier wrote:
> >  
> >>Jens:
> >>
> >>This  pi/4-DQPSK.   That means
> >>
> >>New-symbol  *  (complex conjugate (Old-symbol)) is   pi/4 modulo pi/2.  
> >>Are you taking this into account on both the transmitter and the 
> >>receiver and it in all of the bins before your inverse fft provides the 
> >>time domain signal to transmit would be my best best.
> >>
> >>I will look at it more carefully today if I have time but this question 
> >>would be the first one I would ask the code to tell me.
> >>
> >>Bob
> >>
> >>
> >>
> >>Jens Elsner wrote:
> >>    
> >>>Here is my little problem in animated frames.
> >>>
> >>>What you are seeing is already DQPSK. Displayed is the signal
> >>>constellation for one OFDM frame, 75 symbols. The code was posted 
> >>>earlier.
> >>>
> >>>The problem: The (differential!) signal constellation has a constant 
> >>>phase offset for each OFDM symbol. Why? What to do about it?
> >>>
> >>>Jens
> >>> 
> >>>      
> >>-- 
> >>AMSAT VP Engineering. Member: ARRL, AMSAT-DL, TAPR, Packrats,
> >>NJQRP/AMQRP, QRP ARCI, QCWA, FRC. ARRL SDR Wrk Grp Chairman
> >>Laziness is the number one inspiration for ingenuity.  Guilty as charged!
> >>
> >>    
> >
> >  
> 
> 
> -- 
> AMSAT VP Engineering. Member: ARRL, AMSAT-DL, TAPR, Packrats,
> NJQRP/AMQRP, QRP ARCI, QCWA, FRC. ARRL SDR Wrk Grp Chairman
> Laziness is the number one inspiration for ingenuity.  Guilty as charged!
>