Equalization and detection in nonlinear storage channels with signal-dependent noise

As capacity increases in magnetic storage, signal nonlinearity and correlated, signal-dependent noise begin to inhibit the performance of various detection schemes. To obtain acceptable bit error rates in this environment, one must first understand the limitations of detection schemes currently in use. Using experimentally collected magnetic data, this paper compares the performance of two detection schemes used in magnetic recording: (1) linear equalization followed by sequence detection implemented by the Viterbi algorithm and (2) fixed delay tree search with decision feedback (FDTS/DF). In addition, the performance of these detectors, employing a modified branch metric computation, is determined. The metric incorporates an extra term that proves effective in combatting signal-dependent noise