Comparison between the Least-Squares Lattice and Fast Kalman Algorithms for Adaptive Channel Equalisation

摘要

An exact least-squares lattice algorithm for channel equalisation is derived without using an explicit time update for the equaliser or predictor coefficients. This avoids the need for auxiliary vectors and scalars within the analysis and facilitates a theoretical comparison with the corresponding Fast Kalman algorithm which is also derived. Both algorithms incorporate a very general flow of statistical estimator which includes the growing or growing-fading memory estimators and the sliding window estimator as special cases.