Effect of guardbands ofdm preamble on crosscorrelation

Started by sophie08 December 12, 2008

I m currently simulating a channel estimation based preamble in OFDM
The ofdm signal is convolved with the channel taps which is in this case 3
taps. The technique is to correlate the received preamble with its local
copy in time domain before FFT. This kind of correlation should give an
impulse as the result and in this matter, these impulses should be the
channel taps.
But there is a problem to catch the desired impulses  since its amplitude
is lower than the correlation sidelobe near the main peak correlation.
This is due to the zeros that padded on both side of the preamble as
guardbands. Since the guardbands are should be there, then this problem
seems difficult to deal with. The guradbands appear in higher amplitude
compare to the other sidelobe correlation.
Is there anyone can explain how to deal with this?

Thank you.