University of Tartu - ©2011 Rafik Chaabouni - Last update: 23.10.2013 14:34
Date: 06/11/2013, Time: 17:00, Location: J. Liivi 2, room 317 (next to the coffee room)
Speaker: Prof. Y.L. Guan (Invited speaker from NTU in Singapore)
Title: Coding and Signal Processing for Tera-bits-per-square-inch Magnetic Recording in Future Hard Disks
Abstract:
Shingled-Write/Two-Dimensional, Heat/Microwave Assisted, and Bit Patterned magnetic recording technologies promise to enable Tera-bits-per-square-inch recording density in future hard disk. However, they are confronted with very challenging channel impairment effects such as 2-dimensional inter-symbol interferences, 2-dimensional colored media noise, and unpredictable insertion/deletion errors, as well as prohibitive computational costs. In this talk, coding and signal processing techniques that address some of these challenges will be presented: 1) Reduced-Complexity Iterative Row-Column BCJR Detector for 2-D ISI Channel: This 2-D channel detector essentially reduces the component row or column detector in the well-known IRCSDFA detector from a 2-D BCJR to a 1-D BCJR. Careful iteration scheduling design between the detector and decoder further maintains performance. 2) Dirty-Paper LDPC Code optimized for 2-D ISI Channel: Instead of using an AWGN-optimized LDPC code on a 2-D ISI channel and trying very hard to equalize the severe ISI, we design dirty-paper LDPC codes directly optimized for the "2-D ISI + 2-D detector" super-channel. Dirty-paper code design in this case is possible because the magnetic channel matrix is fairly deterministic per disk. 3) Embedded Marker Code Scheme (EMCS) for Insertion/Deletion Error Channel: EMCS makes intelligent use of the marker/pinning bits to perform the dual functions of re-synchronization and strengthening the outer LDPC code. Soft synchronization, instead of hard synchronization, further improves performance. 4) 2D Coding for Multi-Track Reading: When the recording density increases beyond a certain limit, writing and reading multiple tracks may make more sense as the adjacent-track interference can now be "absorbed" into the symbols written on multiple tracks. In 3x3 ISI channels, we found that it works very well to encode using non-binary LDPC over GF(4), write every code symbol on two tracks, detect 3 tracks, and perform joint detection-decoding over 4 tracks. These works are done in collaboration with the Data Storage Institute (DSI) of Singapore.
Brief Bio:
Associate Professor Yong Liang GUAN obtained his PhD degree from the Imperial College of London, UK, and Bachelor of Engineering degree with first class honors from the National University of Singapore. He now serves as the Head of the Communication Engineering Division in the School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore. He is also an Adjunct Professor of the University of Electronic Science and Technology of China, and a Faculty Associate of the Institute of Infocomm Research, Agency of Science, Technology and Research, Singapore. His research interests broadly include modulation, coding and signal processing for communication, storage and information security systems. He has published an invited monograph, a book, 3 book chapters, and over 250 journal and conference papers. He is an Associate Editor of the IEEE Transactions on Vehicular Technology, and was an AE of the IEEE Signal Processing Letter. He has led 11 past and present externally funded research projects on advanced wireless communication techniques, ultra wideband radio, coding for 10Tb/in^2 magnetic recording, acoustic telemetry for drilling application etc., with total funding of over SGD 6 million.