Adaptive blind equalization using artificial neural networks

We attempt to use a neural network to solve the channel blind equalization problem. An equalizer is a device which by observing the channel outputs recovers the channel inputs. A blind equalizer does not require any known training sequence for the startup period. We have implemented a blind equalizer using a neural network for channel inputs of (-1, 1). The key to our approach is a three-component error/loss function which controls the hidden layer node output, the final network output and the output layer weight parameters. The neural network is trained using a scaled conjugate gradient method which is faster than the steepest descent algorithms and is free from user-defined parameters. Our method is robust. It makes no assumption about the channel input distribution of channel frequency response and needs fewer taps than conventional blind equalizers. Compared to the popular CMA blind equalizers, our network achieves a significantly lower BER but takes longer to train.