Phase rotation for FFT in OFDM

Hi guys,

I am building a synchroniser for OFDM. I read in a paper that the FFT
window can start anywhere in the Cyclic prefix, & that this only results
in Phase rotation at the output of FFT. 1st, is this right?

2nd, how do we fix these phase rotation? In some papers I read, I think I
am supposed to calculate the angle of the 1st symbol, & rotate all the
other symbols by same degree. But this isnt working?

What am I doing wrong? Is this the correct way to fix phase rotation?

gadha wrote:
> Hi guys,
>
> I am building a synchroniser for OFDM. I read in a paper that the FFT
> window can start anywhere in the Cyclic prefix, & that this only results
> in Phase rotation at the output of FFT. 1st, is this right?
>
correct, see the shifting theorem

> 2nd, how do we fix these phase rotation? In some papers I read, I think I
> am supposed to calculate the angle of the 1st symbol, & rotate all the
> other symbols by same degree. But this isnt working?
>
> What am I doing wrong? Is this the correct way to fix phase rotation?

Obviously not.  Xe(k) = X(k) * exp(j*2*pi*tau*k/N), where tau is your
timing error, N is the fft size (k=0...N-1), Xe(k) is the FFT of the
timing error samples and X(k) is the desired FFT samples.

col