Does the squaring and differentiating TED algorithm work in feedback loop and with nco and interpolater used for correction?
In every paper i read they all used it in feed forward mode .
Reply by Steve Pope●June 24, 20152015-06-24
>On 24.06.2015 7:42, Eshwar varma wrote:
> For Oqpsk my plan is if i can do timing recovery (Carrier Phase
> insensitive). then i can simply delay Q channel by 1/2 symbol duration
> than i want to use Decision directed Pll to remove phase and fine and
> frequency offsets.
One thing to consider is that OQPSK is often (perhaps always) code-speak
for FSK (with appropriate parameters). So if you are doing a literature
search for timing recovery methods, you may wish to look at those
used for FSK, as well as those used for QPSK.
> so i searched google and i found a algorithm which is carrier phase
> insensitive SQUARING AND DIFFERENTIATING BASED TED .
Squaring followed by filtering is very popular. (Perhaps bandpass
filtering is more common than differentiation.)
Good luck.
Steve
Reply by Evgeny Filatov●June 24, 20152015-06-24
On 24.06.2015 7:42, Eshwar varma wrote:
> HI all ,
>
> For Oqpsk my plan is if i can do timing recovery (Carrier Phase insensitive). then i can simply delay Q channel by 1/2 symbol duration than i want to use Decision directed Pll to remove phase and fine and frequency offsets.
>
> For Qpsk i used Gardener TED algorithm .I read some literature on it they said even if we apply a signal which is delayed by 1/tb in q channel it will introduce timing jitter if there is any phase offset in carrier frequency (so, IT IS PHASE SENSITIVE).
>
> so in my plan as i want to use timing sync before Fine frequency sync .I should not use Gardener TED algorithm . so i searched google and i found a algorithm which is carrier phase insensitive SQUARING AND DIFFERENTIATING BASED TED .
>
> LINK :http://www.4shared.com/office/HBCbR8a2ce/Timing_Error_Detector_for_OQPS.html
>
> With this type of TED can i use NCO and parabolic interpolater to correct IT ???
>
> thanks in advance....
>
Hi Eshwar,
Yes, you can. NCO and interpolator perform the function of a resampler
(i.e., changing the input sample rate to whatever output sample rate you
need, like 4 samples per symbol). The resampler utilizes timing error
information produced by the TED, but in other respect it works
independently from the TED. So, yes.
Moreover, there are other options to build a resampler, like using
multirate signal processing (see 2001 Harris "Multirate Digital Filters
for Symbol Timing Synchronization in Software Defined Radios"). For some
reason I don't understand, at comp.dsp a multirate resampler is
considered the mainstream option.
Evgeny.
Reply by Eshwar varma●June 24, 20152015-06-24
HI all ,
For Oqpsk my plan is if i can do timing recovery (Carrier Phase insensitive). then i can simply delay Q channel by 1/2 symbol duration than i want to use Decision directed Pll to remove phase and fine and frequency offsets.
For Qpsk i used Gardener TED algorithm .I read some literature on it they said even if we apply a signal which is delayed by 1/tb in q channel it will introduce timing jitter if there is any phase offset in carrier frequency (so, IT IS PHASE SENSITIVE).
so in my plan as i want to use timing sync before Fine frequency sync .I should not use Gardener TED algorithm . so i searched google and i found a algorithm which is carrier phase insensitive SQUARING AND DIFFERENTIATING BASED TED .
LINK :http://www.4shared.com/office/HBCbR8a2ce/Timing_Error_Detector_for_OQPS.html
With this type of TED can i use NCO and parabolic interpolater to correct IT ???
thanks in advance....
Reply by Eric Jacobsen●May 16, 20152015-05-16
On Fri, 15 May 2015 23:04:06 -0700 (PDT), Eshwar varma
<eshwar93.india@gmail.com> wrote:
>Hi all ,
> I implement qpsk demodulator on National instruments Fpga . Now i want to Demodulate Oqpsk signal . As the difference between qpsk and oqpsk is only the delay of one bit period in q channel.
Just to clarify, the delay between the I and Q channels is one-half
symbol period for OQPSK..
>Q1. Can a qpsk demodulator with some changes demodulate oqpsk data ?
Yes.
>Q2. If it works , what changes i should make ?
> flow of qpsk demodulater what i made..
> 1. frequency shifter (input from step 4)
> 2. Matched filter
> 3. AGC
> 4. coarse frequency offset estimator (Rife and broostyn algorithm).
> 5. Timing recovery (Gardener)
> 6. DDPLL
> 7. symbol demapping
>
> what are the stages to be modified?
The main thing that gets weird with OQPSK is that the timing and phase
synchronization are no longer independent. If the carrier phase
isn't correct, the timing loop cannot lock, and this is different from
a QPSK system. For QPSK the usual acquisition sequence for a
continuous signal is for the timing loop to lock, since it can lock
regardless of carrier phase, and once timing is locked the carrier
loop can then lock the phase since it has proper symbol timing.
In an OQPSK system if the phase rotates ninety degrees the timing loop
wants to shift a half-symbol. So if there is a carrier frequency
offset and the constellation is rotating quickly, the timing loop
essentially cannot lock.
There is also an additional ambiguity in that which two symbol halves
go together switches, so that whatever phase ambiguity the demodulator
eventually locks on to, you may also have to eventually sort out which
bits from I go with which bits from Q, since they are no longer
aligned.
Once locked, however, an OQPSK demodulator is essentially a QPSK
demodulator that shifts one channel a half-symbol somewhere before the
slicer. The only real trick is how to get it into a locked
condition, and there are many ways to do that. The simple way, if
you have a lot of time, is to sweep the frequency very slowly. When
the frequency offset is low the demod can lock when the phase lands on
a happy place for the timing and phase loops to lock.
If you have a system with a preamble or acquisition sequence, then
things may get much easier since the acquisition and ambiguitiy
problems can be solved by processing the preamble. After that the
demodulator is just a QPSK demodulator with the appropriate
half-symbol channel shift.
>Q3. suggest me some papers ?
>
>
>Thanks in advance...
> Hi all ,
> I implement qpsk demodulator on National instruments Fpga . Now i want to Demodulate Oqpsk signal . As the difference between qpsk and oqpsk is only the delay of one bit period in q channel.
>
> Q1. Can a qpsk demodulator with some changes demodulate oqpsk data ?
>
> Q2. If it works , what changes i should make ?
> flow of qpsk demodulater what i made..
> 1. frequency shifter (input from step 4)
> 2. Matched filter
> 3. AGC
> 4. coarse frequency offset estimator (Rife and broostyn algorithm).
> 5. Timing recovery (Gardener)
> 6. DDPLL
> 7. symbol demapping
>
> what are the stages to be modified?
>
> Q3. suggest me some papers ?
>
>
> Thanks in advance...
>
Hi, Eshwar.
Until an engineer with hands-on experience of building these systems
hopefully appears in this thread, I suggest you to try to become the
local OQPSK expert in few quick and easy steps:
(1) Open the books "Synchronization Techniques for Digital Receivers" by
Mengali and D'Andrea and "Digital Communications: A Discrete Time
Approach" by Rice.
(2) Browse the contents of the books, and carefully study every part
whose title mentions either "OQPSK", "offset formats" or "offset QPSK".
You would be surprised with how effective that would be!
(3) While reading, note any references to articles which sound
interesting. After you are done with the books, browse any articles you
like.
Best regards,
Evgeny.
Reply by Eshwar varma●May 16, 20152015-05-16
Hi all ,
I implement qpsk demodulator on National instruments Fpga . Now i want to Demodulate Oqpsk signal . As the difference between qpsk and oqpsk is only the delay of one bit period in q channel.
Q1. Can a qpsk demodulator with some changes demodulate oqpsk data ?
Q2. If it works , what changes i should make ?
flow of qpsk demodulater what i made..
1. frequency shifter (input from step 4)
2. Matched filter
3. AGC
4. coarse frequency offset estimator (Rife and broostyn algorithm).
5. Timing recovery (Gardener)
6. DDPLL
7. symbol demapping
what are the stages to be modified?
Q3. suggest me some papers ?
Thanks in advance...