Tap-partitioned Kalman and LMS chip-level equalization for a DS-CDMA receiver

Abstract

We present a chip-level equalization technique for a DS-CDMA forward link receiver. The proposed equalization algorithm estimates the dominant taps of equalizer, and then partitions the equalizer tap into a Kalman partition and LMS partition according to the dominant power criteria. The tap positions assigned as the Kalman partition are updated using the reduced-complexity Kalman filtering adaptive algorithm, and the tap positions assigned as the LMS partition are updated using the LMS adaptive algorithm. The proposed tap-partitioned, chip-level equalizer exhibits fast convergence speed and is less complex than the conventional Kalman filtering-based equalization algorithm. In addition, while the performance of the conventional reduced-rank chip-equalizer degraded significantly when the tap selection was not optimal, that of the proposed algorithm is robust even when the tap partitioning is not ideal.