Concatenated coding and iterative decoding for magnetic and optical recording

With the increasing need for higher data rates and greater storage densities in recording systems, digital signal processing and coding techniques are areas in which significant gains may be achieved. Currently, digital signal processing techniques for magnetic recording systems involve partial response maximum-likelihood (PRML) sequence detection and practical optical recording systems using PRML are being explored. In addition, recording channels are input constrained and modulation codes that account for this are available. Advances have been made to these types of codes to give them error correcting capabilities by way of increasing minimum distances in output code sequences.; In 1993, parallel concatenation of convolutional codes with iterative decoding, or turbo codes were shown to have remarkable near capacity performance on memoryless additive white Gaussian noise (AWGN) channels. Instead of directly increasing minimum distance, the codes eliminate or decrease the number of low weight codewords through the use of an interleaver. They have been successfully applied to many communication systems including satellite, wireless and packet data systems. In this dissertation, we show that recording systems also benefit substantially from the powerful nature of concatenated codes and iterative decoding.; The objective of this research is to develop methods of improving PRML magnetic and optical recording systems by applying parallel and serial concatenated codes with iterative decoding. Parallel concatenated codes applied to PR channels will first be studied. A system that incorporates the channel detector of the PR system in the iterative decoding algorithm will be introduced. We will then present a serial concatenated structure that utilizes the rate one recording channel as an inner code. These systems will be evaluated for simple unconstrained magnetic and optical recording channel models. Since recording channels are input constrained, a turbo coded system that incorporates an RLL code will be developed for a more realistic optical recording channel model. Clarity of how information is shared within the decoding algorithms, channel models, simulation performance, and analysis of these systems will be discussed.