Concatenated coding and equalization for data transmission and storage

The steady growth of data rates in magnetic recording systems is translated into an increase in recording densities. This results in an increase in the nonlinear intersymbol interference (ISI), which degrades the reliability of the retrieved data. The resulting ISI comprises both precursor and postcursor terms. In this dissertation, we introduce new equalization and coding techniques that are superior to previously known techniques. In the equalization area, we propose two new nonlinear equalizers for mitigating the effects of nonlinear ISI in magnetic recording systems. These equalizers, which we call RAM-Search algorithms, have been devised to cancel both precursor and postcursor (nonlinear) ISI. They were tested against two nonlinear models: a simple partial-erasure model for the nonlinear write process, and a third-order polynomial model for the nonlinear read process. A performance improvement of about 2.5 dB at a bit error rate of P(b) = 10⁻⁵ has been shown possible via simulations. The efficacy of these equalizers has also been demonstrated on captured spinstand signals which suffer from a significant amount of nonlinear ISI. In the coding area, we study the serial concatenation of a single convolutional code (CC) and a precoded partial response (PR) channel separated by an interleaver, and is decoded using two iterative decoders that implement the it a posteriori probability (APP) algorithm. One of the weaknesses of this scheme is that its error rate curves tend to hit an error rate floor near P(b) = 10⁻⁶. The level of the floors have been found to depend on the interleaver and precoder (for a given outer CC and a PR channel). In this regard, we discuss analytical techniques that lead to the joint optimization of interleaver and precoder for a fixed outer CC and a given PR channel, where optimality is in the sense of achieving the lowest floor. Such techniques result in lowering the error rate floor significantly at no additional decoder complexity. Additionally, the soft-output Viterbi algorithm (SOVA), which is a suboptimal version of the APP algorithm, is derived for PR channels. We also introduce a slight modification of the SOVA so that it works for arbitrarily high code rates