giny
Newbie level 1
Hi everyone,
I do have problems implementing a 2x1D Wiener-Filter as channel estimator for an OFDM-System. I use the concept introduced in the following paper:
**broken link removed**
First of all I want to make shure I understood the problem. Pls correct me if I am wrong:
What I don't understand:
Can anyone tell me what I missunderstood and push me into the right direction. Or give me a hint how to implement such a channel estimator?
Thanks in advance.
P. S.
\[
k' = sample from Pilot in time domain\\
l' = sample from Pilot in freq. domain\\
k'' = sample from Pilot subset in time domain\\
l'' = sample from Pilot subset in freq. domain\\
\]
I do have problems implementing a 2x1D Wiener-Filter as channel estimator for an OFDM-System. I use the concept introduced in the following paper:
**broken link removed**
First of all I want to make shure I understood the problem. Pls correct me if I am wrong:
- If we do not know the channel well, we can not compute the cross-covariance vector of the channel, but we can use equation (15) assuming a worst case scenario for doppler and delay spread.
- Also the auto-covariance matrix could be calculated with equation (16)
- After that we can minimize the cost-function \[J\] (eq. (4)) or use eq. (3)
- Having computed the weight coefficients we can use eq. (2) with the Least-Square estimation \[r(k',l')\] of the channel at the Pilot Positions to compute a estimation matrix for the hole channel at every location.
What I don't understand:
- Using equation (15): I do get a vector for each carrier location of my OFDM-Symbol, thats bigger than what I expect what \[N_{tap}\] should be?
- Why do we use \[k', k''\] resp \[l',l''\] in eq (16) How do I get the auto-covariance matrix \[N_{tap}\] × \[N_{tap}\]?
Can anyone tell me what I missunderstood and push me into the right direction. Or give me a hint how to implement such a channel estimator?
Thanks in advance.
P. S.
\[
k' = sample from Pilot in time domain\\
l' = sample from Pilot in freq. domain\\
k'' = sample from Pilot subset in time domain\\
l'' = sample from Pilot subset in freq. domain\\
\]