Iterative decoding for magnetic recording channels.

Abstract

The success of turbo codes indicates that performance close to the Shannon limit may be achieved by iterative decoding. This has in turn stimulated interest in the performance of iterative detection for partial-response channels, which has been an active research area since 1999. In this dissertation, the performance of serially concatenated recording systems is investigated by computer simulations as well as experimentally. The experimental results show that the iterative detection algorithm is not sensitive to channel nonlinearities and the turbo coded partial-response channel is substantially better than partial-response maximum-likelihood channels. The classical iterative decoding algorithm was originally designed for additive white Gaussian noise channels. This dissertation shows that the performance of iterative detection can be significantly improved by considering the noise correlation of the magnetic recording channel. The idea is to iteratively estimate the correlated noise sequence at each iteration. To take advantage of the noise estimate, two prediction techniques were proposed, and the corresponding systems were named noise predictive turbo systems. These noise predictive turbo systems can be generalized to other detector architectures for magnetic recording channels straightforwardly.