Information Theory Analysis of Multilevel Flash Memory Part 1. Channel Model and Random Coding Bounds
Abstract
Introduction: Flash memory is one of the fastest growing segments in the global semiconductor industry. Due to the steadily increasing recording density, high read/write speed, low power consumption and long service life, flash memory is used to store data in a very wide field of applications. Increasing the recording density achieved by decreasing the physical size of the memory cell, and the increasing number of states of a memory cell lead to a decrease in data storage reliability, which requires the use of error-correcting coding. Purpose: We study theoretically achievable values for the main parameters of error-correcting coding in a multilevel NAND flash memory model, depending on the recording/storage conditions. Results: Theoretically achievable coding rate limits have been obtained within which it is possible to build reliable memory for one of possible flash memory models treating a memory page as a system with independent multilevel cells. Random coding bounds have been derived for this model, and some numerical results presented. Using these bounds, we establish certain trade-off relationships between the achievable coding rate and the key memory parameters such as the number of program/erase cycles and the data retention time. Gaussian approximation for the channel model has also been considered. This approximation in some cases is sufficiently accurate to get an exact expression for the bound on the channel cutoff rate RPublished
2016-04-21
How to Cite
Trofimov, A., & Taubin, F. (2016). Information Theory Analysis of Multilevel Flash Memory Part 1. Channel Model and Random Coding Bounds. Information and Control Systems, (2), 56-67. https://doi.org/10.15217/issn1684-8853.2016.2.56
Issue
Section
Information coding and transmission