You are not logged in.
Hello,
We were going through the "ofdm_txrx_mimo" model. We had some problems trying to understand some points in channel estimation. So we would like to know:
(1) Is Channel Estimation done using zero forcing?
(2) Is it done using the training sequence only or the subcarrier pilots too ?
(3) Is the channel estimation h(n) for each sub-carrier or interpolation is done for all these sub-carriers (Assuming Flat Fading) ?
(4) Is the channel estimation done all over the packet (using some king of feedback)or only on the training sequence (and pilots) ?
(5) What is the function of the adder in the "ofdm_txrx_mimo -> ofdm rx mimo -> fft & chan Est -> Channel Estimation -> H_A-A Estimate Buffer -> I Estimate Buffer".
Thank you in advance,
Offline
(1) Yes, it is zero forcing.
(2) I believe the pilots are only used for carrier frequency offset (CFO) recovery at the moment. The training is based only on the training sequence.
(3) A single complex channel coefficient is calculated for each subcarrier. The training sequence insures that the the channel is sounded for every subcarrier
(4) Only training. We've had success modifying the one-shot system into an adaptive scheme that corrects the channel coefficients over the course of the packet, but we have not integrated that work into the reference design.
(5) I'll check with the designer of the block and reply later.
Offline
Thanks, this was helpful, we were able to move forward in the design.
Offline
5) That block takes the running sum of multiple estimates per subcarrier. The equalizer uses a single estimate per subcarrier, so the channel estimator takes the sum of the estimate derived for each subcarrier from each training symbol. Think of it as an average estimate, without the division (the division is handled later, by a scaling stage after the equalizer).
Offline