On the Distribution of Bit Errors in NAND Flash Memories




Abstract: In this talk, we present a new method for modeling the distribution bit errors in NAND flash memories. Specifically, we show that the distribution of bit errors in a block of N bits can be modeled as a composite probability mass function (pmf). With a more detailed discussion of the model, we will show some comparisons of the modeled and measured pmf for two different flash memories under varying conditions. The simulation results on this novel model with various ECCs match the tests on NAND.

Bio: Xiaojie Zhang received Ph.D. degree in EE from UCSD in 2012. Previous, he was with Samsung Electronics for 3 years, and with Tidal systems in 2013. Since 2014, he has been with CNEX Labs as algorithm development manager on non-volatile system, focusing on data integrity, error handling, media characterization, and FTL algorithm design. His research interests include wireless communication, error correction coding theory and its applications, and non-volatile memory system design.