Equalization and Detection for High-Density Magnetic Recording Channels Investigators: Nick M. Zayed and L. Richard Carley This project is concerned with developing new signal processing methods that can increase the density of information stored on magnetic disks or tape. The magnetic recording channel is another type of commumications channel and many of the same techniques used in digital communications apply. The primary obstacles to increasing recording density, from a signal processing perspective, are nonlinear intersymbol interference (ISI), correlated, nonstationary noise and, simply, the extensive length of the ISI. At low and moderate densities, the magnetic recording channel's readback waveform can be well approximated by a linear superposition of isolated pulses. The noise present in the channel is dominated by the head and read-side electronics and can be characterized as being a stationary, white Gaussian process. In this environment, traditional linear equalization performs well and optimal error rates are achieved using Maximum Likelihood Sequence Detection (MLSD) typically implemented by the Viterbi algorithm. However, as linear densities increase, the channel displays significant nonlinearity and correlated, signal-dependent (nonstationary) media noise becomes important. In this environment, linear equalization is less effective and the performance of Viterbi detection is degraded. To devise equalization and detection schemes that can handle these high-density anomalies, careful analysis of channel nonlinearities and noise characteristics, is required. This analysis will be carried out using real magnetic data collected from a spin stand equipped with several different head-media combinations. In addition to devising system-level requirements for high density recording channels, implementation strategies (IC design) will also be examined. For more detailed information, click here
	research | people | publications | links | home