Viterbi detection of matched spectral null codes for PR4 channels
L. Fredrickson · 1992. Digests of Intermag. International Magnetics Conference · 1992
Partial response systems [1] have the potential for increasing the storage density of magnetic recording systems employing Viterbi detection [2]. Class-IV partial response (PR4) is an appropriate target for magnetic channels at moderate densities. In [3], it is shown that use of a modulation code which has spectral nulls matching the nulls of the channel transfer polynomial can result in significant coding gains over the baseline partial response channel. In this study, the performance of three Dc and Nyquist free codes for the PR4 channel is determined by simulation. The three codes under consideration are all of rate 3/4, and are contained in [3][4][5]. Baseline channel performance is established using interleaved detectors for the PR4 channel. The channel is modeled as an ideal PR4 channel corrupted by colored noise, as would be produced by a zero forcing equalizer applied to a Lorentzian channel at various values of S = pw50/T. In Figure 1, the injected noise spectrum is shown as a function of normalized frequency for various values of S. In Figure 2, theoretical curves are shown for the byte error rates of the baseline and matched spectral null detectors of ideal PR4 with white noise, while simulated byte error rates are shown as scatter points for various values of S, for the code of [3]. Generally, these codes are employed in systems with an outer error correcting code operating on decoded bytes. As an alternative to the detector of [3], an enhanced baseline detector which flags detected code violations for erasure correction in a Reed Solomon code is considered. In Figure 3, a theoretical curve is shown for the ratio of undetectable to detectable errors for ideal PR4 with white noise, while simulated ratios of undetected byte errors to detected byte errors are shown as scatter points for various values of S for the code of [5].