Schemes for equalisation of communication channels with nonlinear impairments

Two novel techniques are described for equalisation of nonlinear intersymbol interference. One is a symbol-by-symbol-type technique and is built around a conventional decision-feedback equaliser by embedding a neural radial basis-function network before the slicer. The other is a maximum-likelihood method exploiting the classification nature of an equaliser. A Viterbi-minimisation approach is employed which circumvents the need for an explicit channel estimation.