Search

In this work, we consider a perpendicular magnetic recording channel where the data is stored perpendicular to the plane of magnetic media. The perpendicular magnetic recording provides significant aerial density gain compared to the conventional longitudinal magnetic recording. This gain comes at the cost of adding a baseline wander noise which degrades the system performance and affects the data integrity. In this paper, we propose a model for the baseline wander noise. The parameters of the proposed model are computed by posing the problem as minimizing the mean square error. We demonstrate that this problem can be appropriately simplified and posed as finding a least squares problem. The proposed model on the baseline wander noise requires the knowledge of information bits to estimate the noise but this information is unavailable at the time of decoding. We propose a novel technique to cancel the baseline wander effect on the Viterbi trellis. This cancelation is done on the fly in the trellis and obviates the need of the bits information yet provides the performance gain close to the genie performance. Simulation results corroborate the analysis of the proposed algorithm and demonstrate that the proposed algorithm provides a significant performance gain.

The enormous growth of consumer electronic industry is driving the need for achieving ultra-high areal storage densities. Shingled magnetic recording (SMR) is a promising high density storage technique where shingles are created by overlapping tracks for increasing the track per inch (TPI) and achieving higher areal densities. High TPI values, however, introduce sidetrack interference in the equalized output of the center track sensed through the read head. The problem of removing interference from the center track signal can be formulated as estimating pulse shapes of sidetracks assuming sidetracks are decodable. In this paper, we first propose a solution using the correlation method for estimating the pulse shape and the phase-misalignment. The correlation method fails when there is a frequency offset present from sidetracks. Furthermore, we devise a solution based on the least mean square (LMS) algorithm which can provide time varying estimates of the sidetrack pulse shapes and remove the interference from the center track signal. Experimental and simulation results which corroborate the analysis of the proposed algorithm, demonstrate that the proposed algorithm outperforms other techniques and gives the minimum bit error rate (BER). The complexity of the proposed algorithm is low and it can be easily implemented on the hardware.