Abstract
A problem of blind estimation of noisy channels in CDMA (Code Division Multiple Access) system is considered. It is shown that noise term can be eliminated by using the second order statistics (covariance matrix) of received data to obtain a noiseless system whose characteristic is the same as that of the original channel. And based on the derived noiseless system, the channel parameter can be estimated by employing oblique projections and the property of spread sequences. Moreover input signals can be estimated by using the minimum mean square error detector. Some simulation results are included.
