ABSTRACT
The application of multilayer perceptrons to equalize time-invariant nonlinear channels with intersymbol interference is described. The multilayer perceptron is used in a decision feedback structure according to a conventional decision feedback equalizer. Simulation results show that the bit error rate performance of the neural network is significantly superior compared with the decision feedback equalizer if nonlinearities influence the transmission. The performance improvement depends on type and strength of the nonlinearity.
