Abstract:
This paper offers an overview of complex equalizers, which combine the structure of a linear transversal filter (LTE) with a neural network. There are presented nonlinear channel and nonlinear device models. There are exposed different equalizers architectures and some of the training algorithms. Simulation results are presented and it is made a comparative study of the complex equalizers performances, from the point of view of signals space partition and processing error, in different conditions of noise. Concluding remarks and further developments are discussed.