Channel estimation in fast fading channels

Accurate radio channel estimation is necessary for efficient wireless communications. This paper proposes a novel Kalman filter algorithm using basis expansion models (BEMs) of the channel gains to accurately estimate fast fading channels. The channel state for finite duration blocks is summarized in terms of BEM coefficients. The dependency between channel blocks is described with a second-order autoregressive model for the time evolution of the BEM coefficients. A Kalman smoother, based on the dynamic model, is used to combine information from measurements made in different blocks. The mean square error (MSE) of the Kalman smoother is near that of the optimal Wiener filter. The computational cost of this new algorithm is on the same order as existing channel estimation algorithms. This paper is robust to variations of the radio channel parameters from the design values.