Recursive blind estimation and equalization of time-varying channel based on parametric model

This paper proposes a new technique for recursive blind equalization of a time-varying IIR communication channel to obtain simultaneous estimation of channel impulse response and input symbols. The received sequences are represented as the output of a noisy non-Gaussian time-varying parametric model. A pseudo maximum likelihood estimation algorithm is proposed for the identification of channel parameters. The blind equalization is implemented by three algorithms: the recursive channel estimation algorithm, the Gaussian-mixture parameter estimation algorithm and the standard Kalman filtering algorithm. The equalization results are good even on low SNR received sequence and fast fading channel.