Solving closely-spaced multipaths via extended Kalman filter in WCDMA downlink receivers

Delay estimation is an important task in wideband CDMA (WCDMA) receivers. This is a very challenging topic when the successive paths are spaced at less than one chip distance (closely-spaced paths). Traditional delay locked loops (DLLs) are not able to estimate closely-spaced path delays with good accuracy. We introduce an extended Kalman filter (EKF) based algorithm for the estimation of closely-spaced path delays in WCDMA environments. The EKF might be applied either at the sample or at symbol level and it can be also used for the estimation of the complex coefficients of the channel multipaths. The model parameters and the sensitivity to the initialization conditions are described for a practical downlink WCDMA scenario, employing root raised cosine pulse shapes. Simulation results show that EKF is fast convergent both when the estimation is done at symbol or at sample level, and the mean delay error is at most one sample for both cases of oversampling factors of 4 and 8. We show also the conditions when the convergence is not guaranteed.