A Linear Support Vector Machine Based Detector for Bit-Patterned Magnetic Recording

Abstract

The demand for high-capacity storage devices for storing digital information is continuously increasing because of the rapid growth in the number of social media users. Alternative magnetic recording technologies, such as bit-patterned magnetic recording (BPMR), have been proposed in parallel with the current perpendicular magnetic recording technology. However, to increase the areal density in BPMR, we unavoidably encounter the problems of two-dimensional (2D) interference and track mis-registration (TMR). Consequently, to solve these problems, we first present the modified soft-information adjuster (SIA) to mitigate the 2D interference and improve the log-likelihood ratios (LLRs) that were initially produced from the conventional detectors. Then, we propose a linear support vector machine (LSVM)-based detector that works with the modified SIA so as to enhance the reliability of LLRs, which can in turn provide better estimated user bits. Simulation results reveal that the proposed system can yield better bit-error rate performance and is more robust to the TMR effect than the conventional system without the LSVM-based detector.