New methods for evaluating equalizer error rate performance

A new method is presented to analytically determine the steady-state error rate performance of finite length linear equalization and decision feedback equalization (DFE). Usually the error rate performance for finite length equalization is bounded based on the mean square error (MSE) or worst case distortion. The exact expression for bit error rate (BER) performance determined assumes perfect channel estimation and is based on determining the pulse shape at the output of the equalizer. The effects of the residual intersymbol interference (ISI) and additive white Gaussian noise (AWGN) are considered. Thus linear equalization schemes can be analytically compared to DFE on the basis of error rate without simulation. The new error rate calculation for DFE can be applied to the Markov model of DFE error propagation to dramatically improve the characterization of DFE error propagation.