Enhancing Signal Processing With Multilayer Perceptron Utilization for Bit-Patterned Magnetic Recording

Abstract

Bit-patterned magnetic recording aims to overcome the superparamagnetic limit and achieve higher areal density in magnetic recording, with improved thermal stability and reduced transition noise. However, the inter-symbol interference and inter-track interference effects remain crucial problems that must be addressed. This work introduces a multilayer perceptron-based equalizer operating in conjunction with an adaptive target scheme. Furthermore, we also propose using a multilayer perceptron-based detector to improve the overall recording performance. We evaluated our enhanced processing methods using regular and staggered bit-island arrangements. The simulation results indicate that the multilayer perceptron-based equalizer achieves lower mean square error values across all signal-to-noise ratio levels when compared to the conventional minimum mean square error equalizer. Additionally, the proposed multilayer perceptron-based detector performs slightly better than the multilayer perceptron-based equalizer. Also, these proposed techniques demonstrate superior bit error rate performance, even when considering bit-island position fluctuations.