Two-dimensional magnetic recording (TDMR) coupled with shingled-magnetic recording (SMR) is one of next generation techniques for increasing the hard disk drive (HDD) capacity up to 10 Tbit/in2 in order to meet the growing demand of mass storage.

We focus on solving the tough problems and challenges on the detection end of TDMR. Since the reader works on the overlapped tracks, which are even narrower than the read head, the channel detector works in an environment of low signal-to-noise ratio (SNR), two-dimensional (2-D) inter-symbol interference (ISI) and colored noise, therefore it requires sophisticated detection techniques to provide reliable data recovery. Given that the complexity of optimal 2-D symbol detection is exponential on the data width, we had to choose suboptimal solutions.

To build our research environment, we use an innovative Voronoi grain based channel model which captures the important features of SMR, such as squeezed tracks, tilted bit cells, 2-D ISI, electronic and media noise, etc. Then we take an in-depth exploration of channel detection techniques on the TDMR channel model. Our approaches extend the conventional 1-D detection techniques, by using a joint-track equalizer to optimize the 2-D partial-response (PR) target followed by the multi-track detector (MTD) for joint detection, or using the inter-track interference (ITI) canceller to estimate and cancel the ITI from side tracks, followed by a standard BCJR detector. We used the single-track detector (STD) for pre-detecting the side tracks to lower the overall complexity. Then we use pattern-dependent noise prediction (PDNP) techniques to linearly predict the noise sample, so as to improve the detection performance under colored media noise, and especially the data dependent jitter noise. The results show that our 2-D detectors provide significant performance gains against the conventional detectors with manageable complexity.