Read/Write Channel Modeling and Two-Dimensional Neural Network Equalization for Two-Dimensional Magnetic Recording

Modeling of Writing Process for Two-Dimensional Magnetic Recording and Performance Evaluation of Two-Dimensional Neural Network Equalizer

“…The discretized granular medium with non-magnetic grain boundaries based on the discrete Voronoi model is made by our modeling method of medium [4]. The average grain size, the normalized grain size dispersion, and the average non-magnetic grain boundary are set to nm, , and nm, respectively.…”

“…3, the influence of JLMN seems to be extremely large. The reproduced waveform from a reader is calculated by the two-dimensional convolution of the media magnetization and the closed-form reader sensitivity function [4]. It is assumed that the width between side shields, the shield gap, the width, and thickness of the magneto-resistive element of the reader are 26 nm, 15 nm, 13 nm, and 2 nm, respectively.…”

“…The read/write (R/W) channel model based on a discrete Voronoi model [2]- [5] is relatively easy to implement in terms of signal processing development for TDMR. We have developed the modeling of a discretized granular medium with non-magnetic grain boundaries [4] and the simple writing process with an artificial clustering method [5] which does not include intergranular exchange fields and magnetostatic interaction fields between grains based on the discrete Voronoi model. In this paper, the simple writing process is modified so that magnetic clusters are naturally formed by including these fields.…”

“…It is also well known that magnetic clusters formed by the intergranular exchange coupling cause an increase in jitter-like medium noise (JLMN) [6]. A two-dimensional neural network equalizer (2D-NNE) designed by a hybrid genetic algorithm (HGA) [4], [5] can mitigate the influences of not only ITI but also JLMN in TDMR on the bit-error rate (BER) performance. However, a high-level noise due to a misequalization at the output of the 2D-NNE may bring the significant performance degradation in the low-density parity-check (LDPC) coding and iterative decoding system.…”

A Study of TDMR Signal Processing Opportunities Based on Quasi-Micromagnetic Simulations

“…The writing and readback timings are assumed to be perfect. The reading process is modeled by a double-shielded reader sensitivity function [6], which is obtained by fitting to sensitivity function generated from the 2-D-finite element method (FEM) for different head and medium geometries based on a calculation method that appeared in [12]. As noted in [4], in contrast to conventional recording systems, the primary source of noise in TDMR comes from the irregular boundaries of grains and the lack of knowledge of these boundaries during the readback process.…”