A Multilayer-Perceptron based Method for Track Misregistration Mitigation in Dual-reader/Two-track Reading BPMR Systems

Bit-patterned magnetic recording (BPMR) is the future hard disk drive technology that is expected to gain an areal density of over 4.0 Terabit-per-square-inch (Tbits/in
²
). However, one of the serious problems is track misregistration (TMR), which easily degrade the overall system’s performance. To ensure that the system’s performance is acceptable; therefore, we present the TMR mitigation method on a dual-reader/two-track reading (DRTR) BPMR system using an artificial neural network (ANN) model and deep learning technique. To estimate TMR levels, the equalized readback signals are directly fed into a proposed TMR estimator that is performed based on a multilayer perceptron (MLP). In the TMR correction process, both the estimated TMR and equalized readback signals are fed to the MLP detector to detect recorded data bits. The simulation results reveal that the utilization of our proposed TMR mitigation method can improve the bit-error-rate performance of the BPMR system when they were compared with the system that uses another mitigation method.