Optimum quantization for signal processing and error correction in NAND flash memory

Abstract

Conventional error correction employing hard-decision decoding algorithms is not sufficient to correct all the bit errors in high density NAND flash memory. Recently, signal processing algorithms as well as soft-decision error correction are widely studied for improving error correcting performance. However, these techniques demand many memory sensing operations and, as a result, can lead to long sensing latency and high access energy. In this paper, we present optimum memory sensing schemes needed for estimation of threshold voltage distribution, cell-to-cell interference cancellation, and soft-decision error correction of NAND flash memory. We show the error performance improvement with each of these algorithms using simulated NAND flash memory.