Soft Information Scheme Based Multilayer Perceptron in Coded Bit-Patterned Media Recording System

Abstract

In order to increase the efficiency of the recording system, both the inter-symbol interference (ISI) and inter-track interference (ITI) effects that impede the increase of an areal density (AD) in an ultra-high magnetic recording system, such as a bit-patterned media recording (BPMR) system, must be addressed. To improve the soft information or log-likelihood ratios (LLRs) obtained from the soft-output Viterbi algorithm (SOVA), we propose to use a multilayer perceptron (MLP), where three different patterns of the obtained soft information are employed as input data of MLP neural network in the training process under the coded recording system, while the recorded bit data sequences are fixed as a target data bit sequence. When compared to magnetic recording systems based on the partial response maximum likelihood (PRML) technique in both one and two-dimensional signal processing systems, the proposed recording system performs better in terms of bit error rate (BER), according to the results of computer simulation experiments.