OFDM Transmit diversity

  Hello,

    Assume you have 2 OFDM transmitters that are simultanously
transmitting different data and the data is somehow coded (ie CDMA).
After IFFT the ofdm symbol of each transmitter , X1 and X2, passes a
freq. selective fading channel:  Y = X1.G1 + X2.G2 + N

My question is at the receiver, before or after the FFT, how to remove
the effect G1 and G2 which are the fading coefficients.

  Thank you
  Shahin

On May 2, 5:35 pm, Shahin <gheitan...@gmail.com> wrote:
>   Hello,
>
>     Assume you have 2 OFDM transmitters that are simultanously
> transmitting different data and the data is somehow coded (ie CDMA).
> After IFFT the ofdm symbol of each transmitter , X1 and X2, passes a
> freq. selective fading channel:  Y = X1.G1 + X2.G2 + N
>
> My question is at the receiver, before or after the FFT, how to remove
> the effect G1 and G2 which are the fading coefficients.
>
>   Thank you
>   Shahin


How about compute the inverse and convolve?  Like zero-forcing
equalization?

Julius

On May 3, 2:21 pm, julius <juli...@gmail.com> wrote:
> On May 2, 5:35 pm,Shahin<gheitan...@gmail.com> wrote:
>
> >   Hello,
>
> >     Assume you have 2OFDMtransmitters that are simultanously
> > transmitting different data and the data is somehow coded (ie CDMA).
> > After IFFT theofdmsymbol of each transmitter , X1 and X2, passes a
> > freq. selective fading channel:  Y = X1.G1 + X2.G2 + N
>
> > My question is at the receiver, before or after the FFT, how to remove
> > the effect G1 and G2 which are the fading coefficients.
>
> >   Thank you
> >  Shahin
>
> How about compute the inverse and convolve?  Like zero-forcing
> equalization?
>
> Julius

 If you mean the Inverse FFT and convolving with the known channel,
computational wise it would be very complex. isn't it?

Shahin

On May 3, 12:16 pm, Shahin <gheitan...@gmail.com> wrote:
> On May 3, 2:21 pm, julius <juli...@gmail.com> wrote:
>
>
>
> > On May 2, 5:35 pm,Shahin<gheitan...@gmail.com> wrote:
>
> > >   Hello,
>
> > >     Assume you have 2OFDMtransmitters that are simultanously
> > > transmitting different data and the data is somehow coded (ie CDMA).
> > > After IFFT theofdmsymbol of each transmitter , X1 and X2, passes a
> > > freq. selective fading channel:  Y = X1.G1 + X2.G2 + N
>
> > > My question is at the receiver, before or after the FFT, how to remove
> > > the effect G1 and G2 which are the fading coefficients.
>
> > >   Thank you
> > >  Shahin
>
> > How about compute the inverse and convolve?  Like zero-forcing
> > equalization?
>
> > Julius
>
>  If you mean the Inverse FFT and convolving with the known channel,
> computational wise it would be very complex. isn't it?
>
> Shahin

Sure, but it's not the only way to do channel estimation, right?
Isn't that one of the big ideas with OFDM?  Make the bins small
enough such that the effective channel in each subchannel
appears to be "flat" in the frequency domain.  Then all you need
to do zero-forcing equalization is just to divide by a (complex)
number, per channel.  Or, if your modulation is constant-modulus,
do a rotation.

If there's no clock offset in the system, then within each subband
you have a multi-antenna system that appears to have been
subjected to non-frequency selective distortion.

What is it that you are asking again?

Julius