Neural approach to turbo equalization

The turbo equalization can improve the receiver performance significantly. The problem is that the complexity of the optimal algorithms increases exponentially with respect to used code and channel taps. We propose a suboptimal turbo equalizer which is based on iterative equalizer for PSK modulated signals and recurrent neural network convolutional decoder. These algorithms were earlier proposed separately to decrease the complexity of the receiver and it seems possible to make them as an analog chip. Here we combine those methods to form a turbo equalizer. The results show that this can be done and that the idea works well.