Turbo codes are applied to magnetic recoding channels by treating the channel as a rate-one convolutional code that requires a soft a posteriori probability (APP) detector for channel inputs. The complexity of conventional APP detectors, such as the BCJR algorithm or the soft-output Viterbi algorithm (SOVA), grows exponentially with the channel memory length. This paper derives a new APP module for binary intersymbol interference (ISI) channels based on minimum mean squared error (MMSE) decision-aided equalization (DAE), whose complexity grows linearly with the channel memory length, and it shows that the MMSE DAE is also optimal by the maximum a posteriori probability (MAP) criterion. The performance of the DAE is analyzed, and an implementable turbo-DAE structure is proposed. The reduction of channel APP detection complexity reaches 95% for a five-tap ISI channel when the DAE is applied. Simulations performed on partial response channels show close to optimum performance for this turbo-DAE structure. Error propagation of the DAE is also studied, and two fixed-delay solutions are proposed based on combining the DAE with the BCJR algorithm.

We show that high rate interleaved parity check (IPC) codes provide significant coding gain on the magnetic recording channel. A reduced complexity detector performing joint channel detection and IPC decoding is introduced. Simulation result shows an SNR gain of 1.6 dB for the E/sup 2/PR4 channel at a bit error rate of 10/sup -5/, using a rate 196/200 4-way interleaved parity check code.

Advances in VLSI technology permit the use of interpolated timing recovery (ITR) as a replacement for the conventional VCO-based phase lock loop. Fully digital ITR has the advantage of lower cost and higher stability. In this paper, we present an interpolated timing recovery scheme that requires almost no oversampling, which should be suitable for high-speed storage channels.