A joint H2–H∞ algorithm for fading channel estimation and symbol detection

This paper proposes a novel recursive method to jointly estimate the Rayleigh fading channel and the transmitted symbols for current wireless communications systems. Starting from the well-known Jakes’ model for the wireless fading channel, an autoregressive (AR) representation is produced by the means of Levinson–Durbin algorithm. This allows for a state-space approach to the channel estimation problem, for which the MMSE solution is provided by the celebrated Kalman filter. When the robustness of the estimator is important, the H∞ algorithm may be employed instead. It is shown that these algorithms can be combined in order to recover the transmitted symbols and to obtain improved estimates for the envelope and phase of the channel. The paper also proposes a blind channel envelope estimator based on the H∞ algorithm.