A Study of Neural Network Equalization in Perpendicular Magnetic Recording Channel with MR Nonlinear Distortion

Abstract

The neural network equalization in perpendicular magnetic recording channel with jitter-like media noise and MR nonlinear distortion is studied. A method of simplification of neural network using a hybrid genetic algorithm (GA) is proposed. Then, the bit error rate performance for PR2ML system with a simplified neural network equalizer is obtained and compare with those of conventional neural network and transversal filter equalizers. The results show that the SNR improvements of PR2ML system with simplified neural network equalizer over PR2ML system with conventional neural network equalizer and transversal filter equalizer are about 1.0 and 1.5dB, respectively.