Stochastic Systems Group
Home Research Group Members Programs  
Demos Calendar Publications Mission Statement Alumni

SSG Seminar Abstract


Graphical Models and Decoding for Two-Dimensional Intersymbol Interference Channels

Naveen Singla
Washington University


Data storage systems have relied primarily on designs based on storing data on tracks. On magnetic media, input data are encoded and stored as flux reversals on tracks, with decoding being based on standard algorithms such as the Viterbi algorithm. Optical-disc recording uses a similar storage paradigm. As data densities increase, fundamental limits for recording on tracks are approached, and alternative data storage technologies must be considered. Research in data storage systems is shifting towards using two-dimensional storage paradigms. For such paradigms the intersymbol interference is also two-dimensional. This interference invalidates assumptions in the Viterbi and related algorithms necessitating novel decoding strategies. We describe several approaches for the design of encoding and decoding for data storage systems that have two-dimensional (2D) intersymbol interference (ISI). These approaches are based on performing message-passing on different graphical models representing the underlying system. We consider kernel-based and trellis-based graphical models and investigate the performance of belief propagation (BP) and belief revision on the same. The 2D ISI induces short cycles in the graphs which degrade the performance of BP. Alternate message-passing schedules are proposed based on the idea of ordered-subsets (from imaging) to reduce the effect of these cycles. We also show the presence of “channel mismatch” in the BP decoder; the performance of BP can be improved by giving it wrong information about the channel. Potential reasons for the presence of this mismatch and its implications are discussed.


Problems with this site should be emailed to jonesb@mit.edu