Blind identification of channels for discrete multitone (DMT) applications

This paper presents an alternate channel identification approach for discrete multitone (DMT) systems. This method uses a blind algorithm-instead of data-driven algorithms-to identify the channel from the received symbols and knowledge of the statistics of the transmitted symbols. The structure of an ideal DMT system is outlined, and the channel identification algorithm is developed as an extension of the one previously proposed by Tong et al. The degradation in signal-to-noise ratio (SNR), and thus probability of error, due to channel estimation errors is analyzed, and analytic expressions for the multitone SNR as a function of the mean square error (MSE) in channel estimation are derived. Simulation results on a multipath channel with additive white Gaussian noise show good prospects in applying DMT concepts to slowly time-varying channels