From some qualitative materials I read about channel estimation in
OFDM systems, one only need to take out the received pilots, divide
with known pilots and interpolate the samples between pilots, then
divide the chips with these interpolated samples. Is my understanding
correct?
How do I interpolate these complex pilots? Can they be done with simple
lagrange interpolators :?:
there are a number of interpolation methods used in OFDM, namely linear,2nd-order gaussian , lowpass and spline cubic interpolation.
the applied method in each OFDM system depends on system requirments.
From some qualitative materials I read about channel estimation in
OFDM systems, one only need to take out the received pilots, divide
with known pilots and interpolate the samples between pilots, then
divide the chips with these interpolated samples. Is my understanding
correct?
How do I interpolate these complex pilots? Can they be done with simple
lagrange interpolators :?:
Hi,
check out the following book.
MIMO-OFDM WIRELESS COMMUNICATIONS WITH MATLAB by Yong Soo Cho, Jaekwon Kim, Won Young Yang, Chung G. Kang.
It has some matlab code that shows how to insert pilot symbols and do channel estimation using LMS and MSE and channel equalization. You can get the book off the Internet. The examples codes use linear interpolation to estimate the frquency response of the channel.
I have found that polar coordinates interpolation works better, in general. This is fine for MATLAB simulations, but in practice, it will be more complexity since you'll have to use the CORDIC algorithm for fixed-point, real-time implementation.