Turbo-Connected Neural Network Media Noise Cancellation Strategy for Asynchronous Multitrack Detection

Multitrack detection architectures provide throughput and areal density gains over the current industry's standard of single-Track detection architectures. One major challenge of multitrack architectures is the complexity of implementing conventional pattern-dependent media noise prediction (PDNP) strategy within the multitrack symbol detector. In this paper we propose a neural network media noise predictor with manageable complexity that iterates with our rotating target (ROTAR) symbol detector in the turbo equalization fashion to predict and cancel the media noise for multitrack detection of asynchronous tracks. We evaluate the proposed detection strategy on a realistic two-dimensional magnetic-recording channel, and find that the proposed solution can effectively mitigate the media noise and therefore can replace the prohibitively complex PDNP solution for multitrack detection.