Dynamic Threshold Schemes for Multi-Level Nonvolatile Memories




Abstract: In NVMs, reading stored data is typically done through the use of fixed thresholds. However, due to voltage drift, this approach results in significant asymmetric errors. To combat this problem, the notion of dynamic thresholding was recently introduced. In this work, we apply dynamic thresholds to multi-level cell (MLC) memories. The proposed scheme has improved performance over fixed thresholds and can be combined with various flavors of error correction codes.

Bio: Frederic Sala is a second-year Ph.D. student in the Department of Electrical Engineering at UCLA, working with Prof. Lara Dolecek. He received his B.S.E. degree in Electrical Engineering from the University of Michigan, Ann Arbor, in 2010. His research interests are in the area of communications and information theory with a focus on channel coding, including application to synchronization channels and non-volatile memories.