Admissibility in blind adaptive channel equalization

A tutorial survey is presented of an open problem, namely, the admissibility of the memory-less-error-function class of blind adaptive channel equalizers. The problem is treated in a manner that stresses control engineering, rather than communications engineering, terminology. A simple, simulated experiment is set up and referred to throughout. It is used to illustrate intersymbol interference and its removal by a cascaded (fixed) equalizer approximating the inverse channel dynamics followed by a nearest-element decision device. Then, a training signal is assumed to be available at the receiver, and a traditional, prediction-error-based adaptive algorithm (i.e. LMS) is applied, and its parameter estimates observed. The two most widely studied adaptive algorithms for blind channel equalization are then considered.<>