A quadratic sigmoid neural decision feedback equalizer for nonlinear digital magnetic recording channels

A new neural equalizer is proposed in order to compensate for intersymbol interference (ISI) and to mitigate nonlinear distortions in digital magnetic recording systems and to strengthen the bit error rate (BER) performance of conventional neural decision feedback equalizer (NDFE). The proposed quadratic sigmoid neural decision feedback equalizer (QSNDFE) uses a quadratic sigmoid function as the activation function of the neural networks. We present some computer simulation results of the BER performance in nonlinear digital magnetic recording channels-the nonlinear Lorentzian channel model and the partial erasure model. The BER performance of QSNDFE is compared to that of conventional decision feedback equalizer (DFE) and conventional NDFE. Also, the characteristics of MSE (mean square error) of each equalizer is analyzed. It is shown that the QSNDFE outperforms both DFE and NDFE