Low-density parity-check coding for high-density magnetic recording systems.

Abstract

Our strategy is to combine advanced signal processing techniques, the core of which is soft-decision iterative channel detection, with powerful low-density parity-check (LDPC) coding techniques.

Magnetic recording channels (MRCs), including both longitudinal and perpendicular ones, are subject to a number of physical impairments, such as electronic/media noise, intersymbol interference (ISI), erasure, and intertrack interference (ITI). These impairments, if not appropriately handled, are barriers to achieving ultra-high densities. The goal of this dissertation is to study the impact of these multiple impairments on system performance, and to develop techniques to mitigate this impact such that the performance is as close to the theoretical limit of the channel as can be achieved by practical and implementable means.

Specifically, the performance of regular LDPC codes on MRCs is first evaluated. Both randomly and structurally constructed codes are considered. Secondly, density evolution is used to analyze and design LDPC codes for MRCs. Results show that better irregular codes can be obtained. Afterwards, this algorithm is modified to include erasures, and erasure detection algorithms are studied. Fourthly, an improved algorithm for LDPC decoding, called signal-to-noise ratio (SNR) mismatch is unveiled. This algorithm may be useful for future practical applications. Finally, a channel detection algorithm for handling ITI in perpendicular recording is optimized, the eventual goal of which is to maximize the attainable track density.