High Density Turbo TDMR Detection With Local Area Influence Probabilistic Model

Abstract

This paper proposes probabilistic message-passing turbo detection algorithms for 2-D magnetic recording that locally estimate magnetic grain interactions with coded data bits, and thus iteratively assist channel decoding. Such local area influence probabilistic (LAIP) techniques are especially effective at densities ranging from about 4 magnetic grains per coded bit (GPB) down to about 1 GPB, where interaction between grains and bits is significant and occasionally a bit will not be written on any grain, and hence will effectively be “overwritten” by bits on surrounding grains. By modeling the interaction among grains, LAIP enables detection of both overwritten bits and wrong-sign non-overwritten bits so that their log-likelihood ratios are assigned small magnitudes for overwritten bits and correct signs for non-overwritten bits. Simulation results with a random Voronoi magnetic media model show that the LAIP-based detector can accurately detect both overwritten bits and severely influenced non-overwritten bits, and that higher user densities and lower bit error rates can be achieved compared with our previously proposed generalized belief propagation (GBP) detector presented at the 2015 Magnetic Recording Conference (TMRC 2015). In addition, the LAIP-based detector's computer run time is 10 000 times faster than the GBP-based detector.