Forums

Timing Recovery for Blind Demodulation

Started by McNab October 3, 2006
Hi!

I'm working on a simulation of blind QAM demodulation in Matlab. So far i
have implemented AGC, Carrier Phase tracking loop and a CMA equalizer -
but i have not yet implemented symbol timing recovery.

I have seen numerous references to Band-Edge Timiing Recovery (BETR), wich
seem to do the trick quite well. I just can't seem to find a lot of
information on how it works! I have read "Passband Timing Recovery in an
All-Digital Modem Receiver" by Godard but haven't been able to make it
work. Is there any other descriptions of BETR out there?

Thanks,

McNab 

 


McNab wrote:
> Hi! > > I'm working on a simulation of blind QAM demodulation in Matlab. So far i > have implemented AGC, Carrier Phase tracking loop and a CMA equalizer - > but i have not yet implemented symbol timing recovery. > > I have seen numerous references to Band-Edge Timiing Recovery (BETR), wich > seem to do the trick quite well. I just can't seem to find a lot of > information on how it works! I have read "Passband Timing Recovery in an > All-Digital Modem Receiver" by Godard but haven't been able to make it > work. Is there any other descriptions of BETR out there? > > Thanks, > > McNab
There isn't a lot in this which is blind equalisation related, because many timing detectors are the first step in the data recovery chain whether the demodulation is blind or trained. Its really a matter of excess bandwidth. If you have lots of excess bandwidth, things like Gardner work. You only need to consider thing like band-edge when the excess bandwidth is somewhat limited, as Gardner falls apart. If the carrier is X, and the symbol rate is Y, the band will be X-Y/2 to X+Y/2, ignoring the excess bandwidth. Consider what happens if you filter tightly around those two band edge frequencies, and slide the baud phase around. Steve

McNab wrote:

> Hi! > > I'm working on a simulation of blind QAM demodulation in Matlab. So far i > have implemented AGC, Carrier Phase tracking loop and a CMA equalizer - > but i have not yet implemented symbol timing recovery.
The key point is the "eyes" of the QAM should be open at least for the small amount at the very beginning. Otherwise the training and locking will take a lot of time or even may never be successful. All of the parameters should be jointly estimated. The equalizer, the carrier PLL and the symbol timing recovery PLL should be done in the way that they don't interact with each other.
> I have seen numerous references to Band-Edge Timiing Recovery (BETR), wich > seem to do the trick quite well. I just can't seem to find a lot of > information on how it works! I have read "Passband Timing Recovery in an > All-Digital Modem Receiver" by Godard but haven't been able to make it > work. Is there any other descriptions of BETR out there? > > Thanks, > > McNab
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
>There isn't a lot in this which is blind equalisation related, because >many timing detectors are the first step in the data recovery chain >whether the demodulation is blind or trained. Its really a matter of >excess bandwidth. > >If you have lots of excess bandwidth, things like Gardner work. You only
>need to consider thing like band-edge when the excess bandwidth is >somewhat limited, as Gardner falls apart. > >If the carrier is X, and the symbol rate is Y, the band will be X-Y/2 to
>X+Y/2, ignoring the excess bandwidth. Consider what happens if you >filter tightly around those two band edge frequencies, and slide the >baud phase around.
Unfortunately I am not guarenteed lots of excess bandwidth, so I guess Gardner is not the solution for me. If you have any references to information on BETR, it would be greatly appreciated. Thanks, McNab
McNab wrote:
>> There isn't a lot in this which is blind equalisation related, because >> many timing detectors are the first step in the data recovery chain >> whether the demodulation is blind or trained. Its really a matter of >> excess bandwidth. >> >> If you have lots of excess bandwidth, things like Gardner work. You only > >> need to consider thing like band-edge when the excess bandwidth is >> somewhat limited, as Gardner falls apart. >> >> If the carrier is X, and the symbol rate is Y, the band will be X-Y/2 to > >> X+Y/2, ignoring the excess bandwidth. Consider what happens if you >> filter tightly around those two band edge frequencies, and slide the >> baud phase around. > > > Unfortunately I am not guarenteed lots of excess bandwidth, so I guess > Gardner is not the solution for me. If you have any references to > information on BETR, it would be greatly appreciated.
I thought I provided the clue. Don't you expect to do any work on this? Steve
>McNab wrote: >>> There isn't a lot in this which is blind equalisation related, because
>>> many timing detectors are the first step in the data recovery chain >>> whether the demodulation is blind or trained. Its really a matter of >>> excess bandwidth. >>> >>> If you have lots of excess bandwidth, things like Gardner work. You
only
>> >>> need to consider thing like band-edge when the excess bandwidth is >>> somewhat limited, as Gardner falls apart. >>> >>> If the carrier is X, and the symbol rate is Y, the band will be X-Y/2
to
>> >>> X+Y/2, ignoring the excess bandwidth. Consider what happens if you >>> filter tightly around those two band edge frequencies, and slide the >>> baud phase around. >> >> >> Unfortunately I am not guarenteed lots of excess bandwidth, so I guess >> Gardner is not the solution for me. If you have any references to >> information on BETR, it would be greatly appreciated. > >I thought I provided the clue. Don't you expect to do any work on this? > >Steve >
You did provide me with a clue, but apparently I must be mathematicly impaired :-). It looked really easy just filtering tightly around the band edges and getting a good spectral line. I have tried this and for some reason I must be doing something wrong. I have tried simulating the approach from "Joint Blind Equalization, Carrier Recovery, and Timing Recovery for High-Order QAM Signal Constellations" by Jablon - Only the part preceding the spectral line filter. Still with no luck. If it's possible, I could use yet another hint... Thanks, NcNab
very good discussions indeed,
                                            really have to appreciate
each and every member of this group for their thoughfull deep insight
answers

i got few papers on this, may be this might be usefull

1.   	Joint blind equalization, carrier recovery and timing recovery
for high-order QAM signal constellations
Jablon, N.K.;
Signal Processing, IEEE Transactions on [see also Acoustics, Speech,
and Signal Processing, IEEE Transactions on]
Volume 40,  Issue 6,  June 1992 Page(s):1383 - 1398
Digital Object Identifier 10.1109/78.139243
Summary: Two existing blind equalization tap update recursions for
64-point and greater QAM (quadrature amplitude modulation) signal
constellations are studied, along with existing and novel carrier and
timing recovery techniques. It is determined that the su.....
	AbstractPlus | Full Text: PDF(1160 KB)    IEEE JNL
Rights and Permissions
	2.  	Joint blind equalization, carrier recovery, and timing recovery
for 64-QAM and 128-QAM signal constellations
Jablon, N.K.;
Communications, 1989. ICC 89, BOSTONICC/89. Conference record. World
Prosperity Through Communications, IEEE International Conference on
11-14 June 1989 Page(s):1043 - 1049 vol.2
Digital Object Identifier 10.1109/ICC.1989.49843
Summary: Two existing blind equalization tap update recursions suitable
for 64-point and greater QAM signal constellations are studied, along
with existing carrier and timing recovery techniques. The superior tap
update recursion is determined to be the const.....
	AbstractPlus | Full Text: PDF(384 KB)    IEEE CNF
Rights and Permissions

>very good discussions indeed, > really have to appreciate >each and every member of this group for their thoughfull deep insight >answers > >i got few papers on this, may be this might be usefull > >1. Joint blind equalization, carrier recovery and timing recovery >for high-order QAM signal constellations >Jablon, N.K.; >Signal Processing, IEEE Transactions on [see also Acoustics, Speech, >and Signal Processing, IEEE Transactions on] >Volume 40, Issue 6, June 1992 Page(s):1383 - 1398 >Digital Object Identifier 10.1109/78.139243 >Summary: Two existing blind equalization tap update recursions for >64-point and greater QAM (quadrature amplitude modulation) signal >constellations are studied, along with existing and novel carrier and >timing recovery techniques. It is determined that the su..... > AbstractPlus | Full Text: PDF(1160 KB) IEEE JNL >Rights and Permissions > 2. Joint blind equalization, carrier recovery, and timing recovery >for 64-QAM and 128-QAM signal constellations >Jablon, N.K.; >Communications, 1989. ICC 89, BOSTONICC/89. Conference record. World >Prosperity Through Communications, IEEE International Conference on >11-14 June 1989 Page(s):1043 - 1049 vol.2 >Digital Object Identifier 10.1109/ICC.1989.49843 >Summary: Two existing blind equalization tap update recursions suitable >for 64-point and greater QAM signal constellations are studied, along >with existing carrier and timing recovery techniques. The superior tap >update recursion is determined to be the const..... > AbstractPlus | Full Text: PDF(384 KB) IEEE CNF >Rights and Permissions > >
I have read reference 1. Unfortunately it pretty much reffered to another paper by Jablon, Farrow and Chou ("Timing Recovery for blind Equalization"). It looks like I have to find that as well - It seems like I'm getting closer to a solution :-) Thanks for all the help... McNab