Numerically Efficient Kalman Filter Based Channel Estimation for OFDM Data Transmission

Abstract

Channel estimation and prediction algorithms are developed for use in broadband OFDM data transmission over non-ideal channels. The scalar complex channel coefficients are described by Gauss–Markov AR models of a given order in state space form to model the channel fading statistics. On this basis, the conventional Kalman filtering and prediction algorithm (CKFPA) is presented as a starting point for further development. A novel numerically stable channel estimation algorithm based on the original KFPA solution, the so-called extended Array UD Covariance Filter (eUD-CF) algorithm, is developed. The accuracy of the eUD-CF estimator is analyzed by the method of computational experimentation. The simulation results demonstrate that the developed algorithm can effectively restrain the CKFPAs instability problem. The aspects of a parallel implementation of the suggested algorithms are also considered.