Bi-directional pattern-dependent noise prediction with LDPC codes for heat-assisted magnetic recording

As areal densities increase, substantial jitter noise is expected in heat-assisted magnetic recording (HAMR). To mitigate the effects of jitter noise, in an earlier work we proposed the bi-directional pattern-dependent noise prediction (BiPDNP) detector, which employs backward linear prediction as well as the conventional forward linear prediction. In this work, we implement BiPDNP in the Bahl-Cocke-Jelinek-Raviv (BCJR) detector, and investigate its performance with low-density parity check (LDPC) codes. For the LDPC coded channel, by combining the BCJR detector with BiPDNP, we observe that a SNR gain of 1 dB (at bit error rate 10^-4 with 30% microtrack jitter) is achieved over the conventional BCJR detector. Further, in HAMR channel modeling the conventional thermal Williams-Comstock (TWC) model with linear relationship for coercivity with temperature is used. In this work, we update the TWC model by using a non-linear relationship for coercivity with temperature.