Algebraic soft-decision decoding techniques for high-density magnetic recording

As areal density and media development in magnetic recording approach the superparamagnetic limit, signal processing for storage systems becomes ever more important because better coding, equalization, and detection techniques can enable systems to operate with a higher reliability at a lower signal-to-noise ratio (SNR). Current fabrication technology makes available computationally powerful semiconductor devices and this permits the consideration of more complicated coding schemes. Turbo and Low Density Parity Check (LDPC) codes can offer performance close to the Shannon Capacity and thus, have generated much interest. Decoding of these codes uses an iterative approach. Industry requires reliable disk-drive performance with bit-error-rates (BERs) less than 10-14. Because analysis of iterative decoding is difficult and computer simulation can not extend to very low BER regions, it remains unknown whether an error floor exits at the target BER in Turbo and LDPC decoding. For this reason, Turbo and LDPC codes have not been immediately incorporated into present read-channel designs. This dissertation explores modern algebraic soft-decision decoding techniques that can replace existing read-channel decoding blocks without large architectural modifications and can further improve the guaranteed system error rate performance.; We first present Guruswami-Sudan (G-S) hard-decision and Koetter-Vardy (K-V) soft-decision list-decoding and consider practical issues in their implementations. We introduce a symbol-based BCJR algorithm to generate accurate symbol-wise reliabilities for input to the soft-decision decoders. We extend the soft-input list-decoder to provide soft outputs and therefore, allow iterations with the symbol-based BCJR. We apply list-decoding to parity-sharing schemes to obtain the benefits of list-decoding an overall higher-rate code, but with only the complexity of decoding the lower-rate component codes. We devise natural architectures for iterative decoding of parity-sharing codes, including iteration both within a code block, as well as between the block decoder and the channel detector. We also develop a software simulator for high-density perpendicular recording based on the microtrack model that enables realistic waveform generation at a rapid speed.; This dissertation is targeted for, but not limited to, magnetic recording and much of its contents can be applied to any general communication system.