OFDM sychronization implementation?

Hi all,
     
    How to achieve synchronization(carrier phase timing) in ofdm? did ofdm
need a pilot? 

Laron wrote:
> Hi all,
>      
>     How to achieve synchronization(carrier phase timing) in ofdm? did ofdm
> need a pilot? 

I haven't worked with any OFDM schemes, but I've reviewed some.  While 
in theory you could get away without using a pilot in OFDM, in practice 
the OFDM protocols I see have pilot tones.  Given the nature of OFDM, 
dedicating a channel to a pilot tone eases the problem of synchronizing 
the receiver considerably, while not wasting too much carrier power.

-- 
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

>Laron wrote:
>> Hi all,
>>      
>>     How to achieve synchronization(carrier phase timing) in ofdm? did
ofdm
>> need a pilot? 
>
>I haven't worked with any OFDM schemes, but I've reviewed some.  While 
>in theory you could get away without using a pilot in OFDM, in practice 
>the OFDM protocols I see have pilot tones.  Given the nature of OFDM, 
>dedicating a channel to a pilot tone eases the problem of synchronizing 
>the receiver considerably, while not wasting too much carrier power.
>
It's easy to understand the synchronization in single carrier
implementation, however,with ofdm receiver, after downconvertion, the
process is first sampling then use fft to get indexes of time domain signal
at one time point? if so,then how to decide the sampling rate and how to
achieve synchronization after fft? 

Just first time touching ofdm, is there any good reference for these detail
implementation? 

Thanks a lot.
B. R.
Laron
      

Could you specify which OFDM system is interested for you? Wimax, LTE, IEEE
802.11 a/g, DVB? Each system has a unique design to implement the 
synchronization. 

Frank
 
>>Laron wrote:
>>> Hi all,
>>>      
>>>     How to achieve synchronization(carrier phase timing) in ofdm? did
>ofdm
>>> need a pilot? 
>>
>>I haven't worked with any OFDM schemes, but I've reviewed some.  While 
>>in theory you could get away without using a pilot in OFDM, in practice 
>>the OFDM protocols I see have pilot tones.  Given the nature of OFDM, 
>>dedicating a channel to a pilot tone eases the problem of synchronizing 
>>the receiver considerably, while not wasting too much carrier power.
>>
>It's easy to understand the synchronization in single carrier
>implementation, however,with ofdm receiver, after downconvertion, the
>process is first sampling then use fft to get indexes of time domain
signal
>at one time point? if so,then how to decide the sampling rate and how to
>achieve synchronization after fft? 
>
>Just first time touching ofdm, is there any good reference for these
detail
>implementation? 
>
>Thanks a lot.
>B. R.
>Laron
>      
>

On Mar 18, 9:15&#4294967295;pm, "Laron" <jason.piker@n_o_s_p_a_m.inbox.com> wrote:
> >Laron wrote:
> >> Hi all,
>
> >> &#4294967295; &#4294967295; How to achieve synchronization(carrier phase timing) in ofdm? did
> ofdm
> >> need a pilot?
>
> >I haven't worked with any OFDM schemes, but I've reviewed some. &#4294967295;While
> >in theory you could get away without using a pilot in OFDM, in practice
> >the OFDM protocols I see have pilot tones. &#4294967295;Given the nature of OFDM,
> >dedicating a channel to a pilot tone eases the problem of synchronizing
> >the receiver considerably, while not wasting too much carrier power.
>
> It's easy to understand the synchronization in single carrier
> implementation, however,with ofdm receiver, after downconvertion, the
> process is first sampling then use fft to get indexes of time domain signal
> at one time point? if so,then how to decide the sampling rate and how to
> achieve synchronization after fft?
>
> Just first time touching ofdm, is there any good reference for these detail
> implementation?
>
> Thanks a lot.
> B. R.
> Laron

One thing you can do is differentially encode from one bin to the
next.

John

>Could you specify which OFDM system is interested for you? Wimax, LTE,
IEEE
>802.11 a/g, DVB? Each system has a unique design to implement the 
>synchronization. 
>
>Frank
> 
>>
Wimax, Thanks.

>>Could you specify which OFDM system is interested for you? Wimax, LTE,
>IEEE
>>802.11 a/g, DVB? Each system has a unique design to implement the 
>>synchronization. 
>>
>>Frank
>> 
>>>
>Wimax, Thanks.
>

To get the timing/Frame synchronization, it is need to perform the preamble
detection in time domain (before FFT) by making use of conjugate symmetry
and repetition property; to get the frequency synchronization, it is need
to correlate the signal with its CP counterpart for each symbol after
timing synchronization.

>>>Could you specify which OFDM system is interested for you? Wimax, LTE,
>>IEEE
>>>802.11 a/g, DVB? Each system has a unique design to implement the 
>>>synchronization. 
>>>
>>>Frank
>>> 
>>>>
>>Wimax, Thanks.
>>
>
>To get the timing/Frame synchronization, it is need to perform the
preamble
>detection in time domain (before FFT) by making use of conjugate symmetry
>and repetition property; to get the frequency synchronization, it is need
>to correlate the signal with its CP counterpart for each symbol after
>timing synchronization.

I' really glad I found this thread as I was just going to create a similar
one. I'm implementing (generic) OFDM system in Matlab and use Schmidl-Cox
synchronization algorithm. The problem I have is that the M(d) (Schmidl-Cox
timing metric) away from the synchronization symbol has very strong peaks,
with the strongest ones following in the data frame just after second
synchronization symbol. Please see a plot of it here:
http://img718.imageshack.us/img718/5716/34169742.jpg
(horizontal is for sample number and vertical is timing metric). The
correlation of the synchronization symbols is seen at the very beginning
and around x=130 and it is exactly equal to 1 (ideal channel) and length 5
samples. In this example two frames were transmitted, each contained 2 SC
synchronization symbols (c1 and c2) and 8 data symbols.

At the moment I'm using 8 QAM4 sub-carriers, cyclic prefix length is 5
samples and nothing else (thus resulting OFDM symbol length is 13 samples).
The synchronization symbol is created as described in SC paper, i.e. points
are chosen from QAM64 constellation.

Thanks for any clues on what might be causing this problem and what can be
done about this.