A WEIGHTED AUTOCORRELATION METHOD FOR DOWNLINK CDMA LMMSE EQUALIZERS

摘要

Receiving downlink CDMA signals in a fast-fading environment is facilitated at higher receiver velocities by updating the block-adaptive linear minimum mean square error (LMMSE) downlink CDMA equalizer. The autocorrelation matrix of the observed data is updated by passing block-wise autocorrelation slides through a-filter. Each autocorrelation slide is an autocorrelation matrix estimated from a short block of observed data over which the channel can be considered constant. This method achieves a reliable estimate for the autocorrelation matrix when the block size must be small to ensure that the block-wise stationarity assumption holds in cases of fast fading channels. In addition, small block sizes make it possible to satisfy the equalizer delay constraint imposed by hardware and certain voice transmission standards such as CDMA2000 1X where demodulated data must be delivered within only several symbol periods of the signal arrival time. Preliminary simulation results obtained under the 1X standard show that the proposed method outperforms the Rake receiver and the ordinary block LMMSE equalizer in the presence of filtering delay. The improvement over the ordinary block LMMSE equalizer: is substantial in cases of high mobility.