MatsuriSola
Newbie level 2
Currently I want to make a MATLAB simulation of LMMSE channel estimation method in OFDM systems.
In LMMSE method (Wiener Filtering) to find the channel estimation we have to find coefficient :
What I want to ask is how to find these Rxx and Rxy matrices.
I've tried many different methods but they won't work at all...
If you happen to know the answer, please help me. Any programming code would also help me a lot.
========================================================================================================================================================================================================
So far this is what I've tried:
For 7 sub-carriers with pilots separated at distance of 2 subcarriers :
1) For Rxx :
Rxx= Rc(0) Rc(2) Rc(4) Rc(6)
Rc(-2) Rc(0) Rc(2) Rc(4)
Rc(-4) Rc(-2) Rc(0) Rc(2)
Rc(-6) Rc(-4) Rc(-2) Rc(0)
If the received pilot symbols is y, then :
Rc(0) = E[y(f)y(f)]
Rc(1) = E[y(f)y(f-1)]
Rc(2) = E[y(f)y(f-2)]
... and so on...
2) For Ryx : almost the same with Ryx, but I cross-correlate by find the expected value of multiplied received pilot symbols and the whole received symbols (including data symbols)
Here's my works so far in MATLAB :
In LMMSE method (Wiener Filtering) to find the channel estimation we have to find coefficient :
c=(Rxx^-1)*Rxy
where Rxx is autocorrelation matrix and Rxy is cross-correlation matrix.What I want to ask is how to find these Rxx and Rxy matrices.
I've tried many different methods but they won't work at all...
If you happen to know the answer, please help me. Any programming code would also help me a lot.
========================================================================================================================================================================================================
So far this is what I've tried:
For 7 sub-carriers with pilots separated at distance of 2 subcarriers :
1) For Rxx :
Rxx= Rc(0) Rc(2) Rc(4) Rc(6)
Rc(-2) Rc(0) Rc(2) Rc(4)
Rc(-4) Rc(-2) Rc(0) Rc(2)
Rc(-6) Rc(-4) Rc(-2) Rc(0)
If the received pilot symbols is y, then :
Rc(0) = E[y(f)y(f)]
Rc(1) = E[y(f)y(f-1)]
Rc(2) = E[y(f)y(f-2)]
... and so on...
2) For Ryx : almost the same with Ryx, but I cross-correlate by find the expected value of multiplied received pilot symbols and the whole received symbols (including data symbols)
Here's my works so far in MATLAB :