Data-dependent noise-predictive maximum likelihood detection (dd-npml) for magnetic recording

Abstract

In one embodiment, a method includes applying one or more whitening filters to an input stream of digitized samples from a magnetic data channel to produce a filtered sequence, performing one or more branch metric calculations to the filtered sequence to produce a branch metric, and applying a multi-state data-dependent noise-predictive maximum likelihood (DD-NPML) detector to the branch metric to produce an output stream. In another embodiment, a multi-channel data storage system includes a head for reading data from a storage medium, logic for applying one or more whitening filters to an input stream of digitized samples from a magnetic data channel to produce a filtered sequence, logic for performing one or more branch metric calculations to the filtered sequence to produce a branch metric, and logic for applying a multi-state DD-NPML detector to the branch metric to produce an output stream. Other systems and methods are described as well.