An MLP-based Equalization for Bit-Patterned Magnetic Recording Systems

Bit-patterned magnetic recording (BPMR) technology is one of the promising solutions poised to replace a current recording technology so as to enlarge storage capacity. However, the inter-track interference (ITI) and inter-symbol interference (ISI) are a major problem in a BPMR system, especially at high areal densities (ADs). Generally, the conventional system utilizes a partial-response maximum-likelihood (PRML) approach to alleviate ITI and ISI at the receiver, which is a combination of an equalizer and a maximum-likelihood sequence detector. In practice, a conventional equalizer and its corresponding target are designed based on a minimum mean-squared error (MMSE) approach, which may not function well when operating at high ADs because of high non-linearity. Because a multi-layer perceptron (MLP) can be used to effectively handle a non-linear problem, we propose the MLP-based equalizer for a BPMR system to combat the effects of ITI and ISI. Simulation results show that the MLP-based equalizer can provide a lower bit-error rate performance than the PRML approach, and it is resilient to the effect of track mis-registration and media noise. Thus, it may be worth employing the MLP-based equalizer in a BPMR system.