Channel correlation regarding OFDM.

Hello,

I have a confusion regarding the channel correlation of an OFDM system.
In the OFDM receiver, after the FFT, the frequency-domain output is given
as:

Y = XH + W

Y,X,H and W are matrices with X being a diagonal matrix with constellation
points.

My question is regarding channel estimation. Usually in MMSE, we need the
channel correlation R_HH ('_' denotes the subscript}.
And it is usually defined as: R_HH = E{HH^T}  (
(H^T denotes the complex conjugate transpose of H)

The problem is, the correlation is in time-domain or frequency-domain?
H is the row matrix of channel gains of each subcarriers and obviously in
frequency-domain. But when we consider R_HH, is it the correlation in
frequency-domain between subcarriers or time-domain correlation between a
delayed channel transfer function?

Thanks a lot guys.

On Dec 14, 10:53 pm, "m26k9" <maduranga.liyan...@gmail.com> wrote:
> Hello,
>
> I have a confusion regarding the channel correlation of an OFDM system.
> In the OFDM receiver, after the FFT, the frequency-domain output is given
> as:
>
> Y = XH + W
>
> Y,X,H and W are matrices with X being a diagonal matrix with constellation
> points.
>
> My question is regarding channel estimation. Usually in MMSE, we need the
> channel correlation R_HH ('_' denotes the subscript}.
> And it is usually defined as: R_HH = E{HH^T}  (
> (H^T denotes the complex conjugate transpose of H)
>
> The problem is, the correlation is in time-domain or frequency-domain?
> H is the row matrix of channel gains of each subcarriers and obviously in
> frequency-domain. But when we consider R_HH, is it the correlation in
> frequency-domain between subcarriers or time-domain correlation between a
> delayed channel transfer function?
>
> Thanks a lot guys.

Seems like a time domain autocorrelation of the channel gains. Its
same as taking the inverse Fourier transform of the Doppler spectrum
at any given subcarrier of the OFDM symbol. Frequency domain
autocorrelation of the channel is the Fourier transform of the delay
power profile/multipath intensity profile.

>
>Seems like a time domain autocorrelation of the channel gains. Its
>same as taking the inverse Fourier transform of the Doppler spectrum
>at any given subcarrier of the OFDM symbol. Frequency domain
>autocorrelation of the channel is the Fourier transform of the delay
>power profile/multipath intensity profile.
>

Thanks a lot buddy.

Still a bit confused on calculating it in Matlab. 
I have created subcarriers gains in both frequency and time axes from a
Rayleigh fading channel. Stuck in there :(
Appreciate the help mate.
Thanks.