A Fuzzy-Based Outer Loop Controller for LMS Algorithm and its Application to Channel Estimation and Carrier Tracking for OFDM

摘要

A new approach to increase the training speed of the least mean square (LMS) algorithm based on the fuzzy rule base (FRB) is presented. The approach contains an outer loop to adjust the step size of the LMS algorithm. This loop is shown to provide a significant improvement to the training performance and channel tracking capability of the channel estimator. Three effects of channel variations are considered, Doppler Shifts, Carrier Frequency Offsets and Sampling Offsets. These effects need to be tracked and corrected for during the channel estimation phase of an Orthogonal Frequency Division Multiplexing (OFDM) receiver. These effects are important limiting factors on system performance, and need to be compensated effectively using a low complexity and robustly implemented algorithm. Thus, for the standards considered in this paper (HIPERLAN/2 and IEEE801.11a), which are OFDM based; the convergence performance of the simple LMS algorithm is enforced by an experience based outer loop controller (FRB). The FRB provides an opportunity to design a training trajectory for the LMS, where stochastic convergence occurs when the channel profile follows Rayleigh fading statistics in the presence of Additive White Gaussian Noise (AWGN).