Data-Dependent Noise-Predictive Filter Calibration in Low SNR Region for the Magnetic Recording Channel
Jianzhong Huang, Lingyan Sun, Seong-Wook Jeong, Haitao Xia, Weijun Tan · IEEE Transactions on Magnetics · 2014
Noise-predictive filters (NPFIRs) are an effective method to improve the system performance when the noise is correlated and data dependent, such as the media noise in the magnetic recording channel. Although the known data can be used to calibrate the NPFIRs during the manufacturing stage and save them to the zone table as the default setting, when it comes to practice, live decision outputs from the channel detector or channel decoder are needed to fine-tune the NPFIR to match the real media noise statistics better. The live decision driven NPFIR calibration works as well as the known data driven mode when the operating signal-to-noise ratio (SNR) is high. However, in the low SNR region, the performance will be degraded due to: 1) the live decision outputs from the channel detector having a low quality and 2) very few codewords having converged after channel decoding so that the NPFIR does not have enough converged codewords to calibrate. In this paper, we propose two schemes to address the low SNR NPFIR calibration issues accordingly. Simulation results are provided and showed that the proposed two schemes are effective techniques to improve the system performance in low SNR region.