Deep Neural Networks based Soft-Information Improvement for Two-head/Two-track Bit-Patterned Magnetic Recording

Abstract

To increase an areal density (AD) of an ultra-high density bit-patterned magnetic recording (BPMR) system, we have previously proposed a track misregistration (TMR) correction method combined with the soft information adjustor (SIA) to cope with the effects of TMR and two-dimensional (2D) interference. However, we found that soft information or log-likelihood ratio (LLR) can be improved to earn better bit-error-rate (BER) performances. In this work; therefore, we propose to use two types of deep neural networks (DNNs), i.e., multi-layer perceptron (MLP) and long short-term memory (LSTM) network with identical parameter magnitude to improve overall system performance. Here, both DNNs are operated with an earlier SIA on a two-head/two-track (2H2T) BPMR system. Numerical results show that our proposed methods can deliver better BER performance over the earlier SIA system at all TMR levels with and without position jitter noises at the AD of 3.0 Terabit per square inch.