Near maximum likelihood multi-track partial response detection for magnetic recording

The paper considers the detection of data from a parallel multi-track magnetic recording medium in the presence of severe adjacent track interference. In this case, the complexity of a full maximum likelihood detector is prohibitive, even for modest numbers of rows. Hence, a reduced complexity detector based on the use of a tree search algorithm (such as the M-algorithm) with a time varying trellis model of the channel is proposed. The complexity of the detector is related to the parameter M, and it is shown that modest values of M can result in good performance, yet should be feasible to implement.