> Seems like many folks like to correlate a copy of what is received
> with a delayed copy of itself, but this method is not robust esp in
> the presence of multipath.


Who told you that? It is easy to see that without other impairments, delay
correlation will cancel channel effect! Write down some equations and you
should see why!

wongchichian@hotmail.com (wong) wrote in message news:<34a4fe15.0309042210.17839aeb@posting.google.com>...
> Hi
> Would like to know how to do frame synchronization in OFDM systems.
> Seems like many folks like to correlate a copy of what is received
> with a delayed copy of itself, but this method is not robust esp in
> the presence of multipath. What is the best way?
> Thanks all.

If you use double correlation method it will improve against
multipath.
In one correlation,
 you calculate cross correlation using known short preambles and find
the
peaks(Fine time sync.)

Another you do auto correlation(Coarse sync.)  of received samples
with its delayed copy.
This one will give plateau. You have take a joint decision point based
on both
correlations.

Look at Cox's Paper.
Regards,
Santosh

Hi
Would like to know how to do frame synchronization in OFDM systems.
Seems like many folks like to correlate a copy of what is received
with a delayed copy of itself, but this method is not robust esp in
the presence of multipath. What is the best way?
Thanks all.