Developments of the conventional nonlinear equaliser

A computer-simulation comparison of the tolerances to additive white Gaussian noise of various developments of a conventional nonlinear (decision-feedback) equaliser is presented for different levels of intersymbol interference in the received signal. The aim of the investigation has been to clarify the relationships between the different techniques. It is shown that, under the appropriate conditions, a useful improvement in tolerance to noise is obtained by means of a modification to the detection process itself, which, in one case, does not change the complexity of the equaliser and, in the other case, slightly increases it, in return for a further improvement in performance. A more substantial improvement in performance is achieved by a simple modification of the reduced-state Viterbi-algorithm detector that uses ranking to select the stored vectors (sequences), the modification having an effect equivalent to the decision-directed cancellation of intersymbol interference (decision feedback) that takes place in a nonlinear equaliser. To exploit the full potential of this system, as well as that of an ideal maximum-likelihood detector, a relatively large delay in detection must be used.