Adaptive Algorithms Performance Analysis for Adaptive Multiuser Detection over Time Varying Communications Channels

Abstract

This paper analyzes the theoretical limits of the Least Mean Square (LMS) and Recursive Least Square (RLS) adaptive algorithms performance for the adaptive CDMA receivers in slowly time varying communication channels in the presence of multi-path. We show that since the RLS involves memory and does not explicitly recognize time varying nature of the channel (time varying "true" weights), it has worse tracking performance then the LMS adaptive algorithm. Thus, the implementation of the RLS algorithm for adaptive CDMA receivers operating in time varying channels is very limited. The LMS algorithm provides better tracking, but still speed of channel change must be slow. In addition, we present simulation results which confirm the theoretical analysis. Show more