Re: optimal training signals for MIMO OFDM channel estimatio
As the name suggests , pilots are used for estimating the channel .
Pilots are sent with known values to both the TX and Rx.
Pilots are sent between data signals . The higher the pilot density
the better the estimation .
If there has to be a faithful reproduction of the signal at the Rx end
then it is imperative to either know/estimate the channel.
For eg let the value of the pilot ( timedomain) is x
let us say the channel impulse response ( how the channel responds to
a signal )= h
Let the received signal is y. plus some noise n
the Tx-Rx equation can be written as
y= hx+n
Considering noise to be modelled as AWGN cn(0.,var)
y is known, x is known hence h = y/x( simplistically put)
if u poll the received signals for pilots then u will be able to derive
a set of values for h ( h1.. hn) ( coeff of a filter)
For equalisation
if u do some sort of interpolation( cubic/ spline) on the values of h
u will get the approximate values of the transmitted data. This is an approximate
method and one needs to go for mmse) for better.