Convergence and tracking of adaptive reduced-rank interference suppression algorithms

Abstract

The convergence and tracking performance of adaptive reduced-rank interference suppression is studied for direct-sequence (DS)-code division multiple access (CDMA) with randomly assigned spreading sequences. We first consider a reduced-rank filter in which the received signal is partially despread before it is applied to a low-rank multi-stage Wiener filter (MSWF). Partial despreading (PD) reduces the computational complexity associated with the MSWF. The large system convergence analysis of least squares adaptive algorithms presented by Xiao and Honig (see Proccedings of CISS'2000, Princeton, NJ, 2000) is used to evaluate the output signal-to-interference plus noise ratio as a function of number of training samples. Our results show that given a sufficient number of training samples, the combined PD-MSWF performs approximately the same as a training-based adaptive MSWF. We then consider the tracking performance of an adaptive low-rank MSWF in the presence of time- and frequency-selective Rayleigh fading. Our results show that the adaptive low-rank MSWF typically gives a significant improvement in coded error rate relative to a full-rank adaptive filter.