An Adaptive Minimum-Frame-Error Rate Detector for Magnetic Recording

Abstract

A magnetic recording read channel has numerous parameters that must be carefully tuned for its best performance; these include not only the equalizer coefficients, but also any parameters inside the soft-output detector, some of which may be pattern-dependent, including signal levels, predictor coefficients, and residual noise variances. Conventional tuning strategies based on a minimum-mean-squared error criterion are not optimal in terms of frame-error rate and ultimately areal density. Here, we propose a strategy for optimizing the parameters with the aim of minimizing the frame-error rate after error-control decoding. The proposed strategy exploits the close connection between the frame-error rate and the gap between the two curves in an extrinsic information transfer (EXIT) chart. A stochastic gradient algorithm applied to a cost function that quantifies this gap leads to our proposed adaptive minimum-frame-error rate (AMFER) algorithm for adapting the equalizer and detector parameters. Numerical results based on a quasi-micromagnetic simulated channel show that the AMFER parameters can reduce the frame-error rate by more than two orders of magnitude, leading to a 7% gain in areal density over conventional minimum mean-square-error (MMSE) parameters.