Blind channel shortening algorithm based on correlation property of multicarrier signals

Abstract

In this paper, we propose a novel blind channel shortening algorithm to update the coefficients of the time domain equalizer (TEQ) in multicarrier system. The TEQ design investigates a second order statistics property of the received multicarrier signal, given by the presence of the cyclic prefix in the source. Here no assumptions about source stationarity and whiteness are needed to formulate the channel shortening criterion. The proposed shortener is implemented as a stochastic gradient descent algorithm and is compared to existing blind algorithms in terms of data bit rate. The simulation results are studied in the context of ADSL system.