The polynomial-based generalized least mean squares estimator for Rician and Rayleigh fading channels

A simplification of a previously developed polynomial-based generalized recursive least squares estimator for fading channels is presented. The instantaneous inverse correlation matrix, required for each iteration of the original algorithm, is replaced by a time-invariant approximation to reduce the computational complexity. Under some conditions, the simplified algorithm is equivalent to the standard least mean squares algorithm and is thus called the polynomial-based generalized least mean squares channel estimator. Like its predecessor, it employs polynomial channel models and does not require a priori knowledge of the channel statistics. Furthermore, the proposed estimator does not need to be augmented to accommodate channels with a non-fading mean component. Therefore unlike statistical approaches such as Kalman filtering, the polynomial-based generalized least mean squares estimator is applicable to both Rician and Rayleigh channels without any re-configuration and increased complexity. Simulated and analytical results illustrate its tracking performance and the improvement over the standard least mean squares algorithm.