Carnegie Mellon University

Data Storage Systems Center

College of Engineering

Machine Learning for HDD Data Detection

Objective: Understanding How Machine Learning Can Bring ADC Gain for HDD.
Feature: DNN and CNN based neural networks for TDMR data recovery.

We have established some of the most fundamental understanding of using neural networks to perform data detection for HDD channels. For medium noise dominated channel, ML exhibits significant BER gain while at par with conventional channel for white Gaussian noise dominated ones. With multi-read setup in TDMR, multi-reader ML channel presents significant advantage over the conventional one, which could bring significant ADC gain for HDDs.

01_HAMR_Holistic_Modeling.png

 

For Sponsors

Related Materials

Project Overview Slides

Related Publications

Y. Qin and J. Zhu, "Deep Neural Network: Data Detection Channel for Hard Disk Drives by Learning," IEEE Transactions on Magnetics, vol. 56, no. 2, pp. 1-8, Feb. 2020.

Related Presentations

Deep Neural Network: Data Detection Channel by Learning. Yuwei Qin. 2019 DSSC Fall Technical Review.

Request your Sponsor Account.
Having trouble accessing content? Contact Chloë Mattingly for assistance.