On May 4, 2:12 am, "nqh" <n...@yahoo.com> wrote:
> >On May 1, 4:11 am, "nqh" <n...@yahoo.com> wrote:
> >> Hi all,
> >> My project is Digiatal Sampling correction Interpolation/Decimation
> for
> >> application in flexible OFDM Modem.
> >> I'm implementing in Matlab the 5_taps optimized interpolator which
> will
> >> improve the
> >> performance for long data OFDM bursts.
> >> I used control parameters basepoint index (m(k)) and fractional
> interval
> >> (mu(k)) (reference to FM Gardners article "Interpolation In Digital
> >> Modems---PART1: Fundamentals"):
>
> >OK, this is your second post with the same question, and nobody
> >answered your first one.  I think it is because not many people have
> >a copy of that reference paper handy.  It's hard to help you if you
> >don't
> >give enough information for average readers like myself to even begin
> >to understand what the variables mean.
>
> >In particular, what the heck is e(m,0)?  What does the 0 mean?
>
> >I am guessing that Gardner assumes in his paper that you have some
> >way of estimating the timing error, and he focuses on what to do with
> >this timing error.  In that case, you need to have a timing error
> >detector
> >(TED).  Is this what you were asking?
>
> >Julius
>
> Dear Julius,
> It's  dificult to give enough information. So I have a general question:
>
> In OFDM system, we define T is the sampling period at the output of IFFT
> (at transmiter), T' is sampling period at receiver, SFO=(T'-T)/T
>
> At receiver, after CP removal, the (m,n)-th sample of the time-domain
> received signal
> is r'(m,n),(include SFO).
> How can we use Interpolation/decimation technique to correct the r'(m,n)
> (to get desied signal r(m.n) without SFO)
> Who can help me to propose the Algorithm ? (assume perfect knowledge
> SFO).
> Thank you very much!
>

If it's difficult to give enough information then it must be even more
difficult to give a useful answer.

What is m and n?  What is SFO?

What is the big picture here?  Are you trying to use Gardner-type
interpolation-based scheme to do time delay adjustment in an OFDM
system?

You should realize that in an OFDM system the effect of time delay
between the Tx and the Rx is phase rotation, which grows linearly
with frequency.  This is because each subcarrier is multiplied by
a complex factor proportional to delay and offset between carrier and
subcarrier frequency.  This is true assuming that the subcarrier tones
are narrow enough.

If that's what you have been asking, please take some time to do
the math see what the effect of a delay is.

Julius

>On May 1, 4:11 am, "nqh" <n...@yahoo.com> wrote:
>> Hi all,
>> My project is Digiatal Sampling correction Interpolation/Decimation
for
>> application in flexible OFDM Modem.
>> I'm implementing in Matlab the 5_taps optimized interpolator which
will
>> improve the
>> performance for long data OFDM bursts.
>> I used control parameters basepoint index (m(k)) and fractional
interval
>> (mu(k)) (reference to FM Gardners article "Interpolation In Digital
>> Modems---PART1: Fundamentals"):
>
>OK, this is your second post with the same question, and nobody
>answered your first one.  I think it is because not many people have
>a copy of that reference paper handy.  It's hard to help you if you
>don't
>give enough information for average readers like myself to even begin
>to understand what the variables mean.
>
>In particular, what the heck is e(m,0)?  What does the 0 mean?
>
>I am guessing that Gardner assumes in his paper that you have some
>way of estimating the timing error, and he focuses on what to do with
>this timing error.  In that case, you need to have a timing error
>detector
>(TED).  Is this what you were asking?
>
>Julius
>
>


Dear Julius,
It's  dificult to give enough information. So I have a general question:

In OFDM system, we define T is the sampling period at the output of IFFT
(at transmiter), T' is sampling period at receiver, SFO=(T'-T)/T

At receiver, after CP removal, the (m,n)-th sample of the time-domain
received signal 
is r'(m,n),(include SFO).
How can we use Interpolation/decimation technique to correct the r'(m,n)
(to get desied signal r(m.n) without SFO)
Who can help me to propose the Algorithm ? (assume perfect knowledge
SFO).
Thank you very much!
  


_____________________________________
Do you know a company who employs DSP engineers?  
Is it already listed at http://dsprelated.com/employers.php ?

On May 1, 4:11 am, "nqh" <n...@yahoo.com> wrote:
> Hi all,
> My project is Digiatal Sampling correction Interpolation/Decimation for
> application in flexible OFDM Modem.
> I'm implementing in Matlab the 5_taps optimized interpolator which will
> improve the
> performance for long data OFDM bursts.
> I used control parameters basepoint index (m(k)) and fractional interval
> (mu(k)) (reference to FM Gardners article "Interpolation In Digital
> Modems---PART1: Fundamentals"):

OK, this is your second post with the same question, and nobody
answered your first one.  I think it is because not many people have
a copy of that reference paper handy.  It's hard to help you if you
don't
give enough information for average readers like myself to even begin
to understand what the variables mean.

In particular, what the heck is e(m,0)?  What does the 0 mean?

I am guessing that Gardner assumes in his paper that you have some
way of estimating the timing error, and he focuses on what to do with
this timing error.  In that case, you need to have a timing error
detector
(TED).  Is this what you were asking?

Julius

Hi all,
My project is Digiatal Sampling correction Interpolation/Decimation for
application in flexible OFDM Modem.
I'm implementing in Matlab the 5_taps optimized interpolator which will
improve the
performance for long data OFDM bursts.
I used control parameters basepoint index (m(k)) and fractional interval
(mu(k)) (reference to FM Gardners article "Interpolation In Digital
Modems---PART1: Fundamentals"):

   m(k)=int[k*Ti/Ts]    (1)
   mu(k)=k*Ti/Ts - m(k) (2) 

But I have a problem as follow:
Define e(m,0) is timing error of m-th symbol: 
e(m,0) = m*(N+Ng)*SFO
where N: FFT size(=64); Ng: guard size (=16) ; SFO=(T'-T)/T (T is sampling
period at the output of IFFT, T' is sampling period at receiver)

How can I join e(m,0) and (1),(2)? (assume perfect knowledge SFO).


Any help is greatly appreaciated.

Regards,

_____________________________________
Do you know a company who employs DSP engineers?  
Is it already listed at http://dsprelated.com/employers.php ?