A noniterative approach for computing linear equaliser tap weights

Abstract

A fast computation method for the tap weights of a linear equaliser is developed. By approximating the channel autocorrelation matrix, which has a Toeplitz structure, by a circulant matrix, the tap weights can be solved via discrete fourier transform (DFT) techniques. For an equaliser with N taps, the approximate tap weights approach the exact tap weights as N/spl rarr//spl infin/. Excess mean-square error and bit error rate (BER) degradation caused by the approximate tap weights are computed for a representative channel.