In magnetic recording channels (MRCs) the readback signal is corrupted by many kinds of impairments, such as electronic noise, media noise, intersymbol interference (ISI), inter-track interference (ITI) and different types of erasures. The growth in demand for the information storage, leads to the continuing pursuit of higher recording density, which enhances the impact of the noise contamination and makes the recovery of the user data from magnetic media more challenging. In this dissertation, we develop advanced signal processing techniques to mitigate these impairments in MRCs.We focus on magnetic recording on perpendicularly magnetized media, from the state-of-the art continuous media to bit-patterned media, which is a possible choice for the next generation of products. We propose novel techniques for soft-input soft-output channel detection, soft iterative decoding of low-density parity-check (LDPC) codes as well as LDPC code designs for MRCs.First we apply the optimal subblock-by-subblock detector (OBBD) to nonbinary LDPC coded perpendicular magnetic recording channels (PMRCs) and derive a symbol-based detector to do the turbo equalization exactly. Second, we propose improved belief-propagation (BP) decoders for both binary and nonbinary LDPC coded PMRCs, which provide significant gains over the standard BP decoder. Third, we introduce novel LDPC code design techniques to construct LDPC codes with fewer short cycles. Performance improvement is achieved by applying the new LDPC codes to PMRCs. Fourth, we do a substantial investigation on Reed-Solomon (RS) plus LDPC coded PMRCs. Finally, we continue our research on bit-patterned magnetic recording (BPMR) channels at extremely high recording densities. A multi-track detection technique is proposed to mitigate the severe ITI in BPMR channels. The multi-track detection with both joint-track and two-dimensional (2D) equalization provide significant performance improvement compared to conventional equalization and detection methods.
展开▼
