The performance of generalized maximum transition run trellis codes

There has been considerable recent activity in the design and implementation of codes for magnetic saturation recording which achieve a 2.2 dB gain on the ideal EEPR4 partial response channel. This paper compares the performance of one such 8/9 rate Generalized Maximum Transition Run (GMTR) code with a normal 16/17 rate RLL(0,k) code. The sensitivity of the comparison is considered for two idealized magnetic dipulse shapes, for two idealized signal sampling techniques, and for two idealized noise environments. Comparison of a blended Lorentzian and Gaussian dipulse to a pure Lorentzian show generally similar gross effects, but the blended dipulse shows the onset of high density effects at lower channel densities. Comparison of the ideal minimum bandwidth discrete time equivalent sampler to the ideal whitened matched filter sampler shows negligible differences. The GMTR code in Additive White Gaussian Noise (AWGN) achieves net coding gains from 0.2 to 1.7 dB on user densities from 2 to 3 depending on the sequence detector target. In a media noise model the GMTR code is shown to achieve significant coding gains with the EEPR4 or MLSD detector targets.