Increasing flash memory lifetime by dynamic voltage allocation for constant mutual information

Abstract

The read channel in Flash memory systems degrades over time because the Fowler-Nordheim tunneling used to apply charge to the floating gate eventually compromises the integrity of the cell because of tunnel oxide degradation. While degradation is commonly measured in the number of program/erase cycles experienced by a cell, the degradation is proportional to the number of electrons forced into the floating gate and later released by the erasing process. By managing the amount of charge written to the floating gate to maintain a constant readchannel mutual information, Flash lifetime can be extended. This paper proposes an overall system approach based on information theory to extend the lifetime of a flash memory device. Using the instantaneous storage capacity of a noisy flash memory channel, our approach allocates the read voltage of flash cell dynamically as it wears out gradually over time. A practical estimation of the instantaneous capacity is also proposed based on soft information via multiple reads of the memory cells.

Using Dynamic Allocation of Write Voltage to Extend Flash Memory Lifetime

“…This paper systematically explains and demonstrates the concept of dynamic voltage allocation, and addresses several issues to support a practical implementation. This paper extends our previous work [22], [23] in the following ways: 1) Adding programming error and cell-to-cell interference to the ground truth channel model; 2) Analyzing DVA's performance when both estimation and scale-factor adjustment algorithms are simplified by using a simple Gaussian model to approximate the underlying (and more complex) ground truth channel; 3) Designing and analyzing DVA specialized to match an even-odd structure for writing to cells; 4) Exploring the improvement obtained by DVA over DTA; 5) Analyzing DVA's performance when both write and read voltage values are restricted to a finite set of available voltages; 6) Analyzing DVA's complexity. This paper begins by introducing DVA in an idealized setting and then removes ideal assumptions about channel knowledge and threshold resolution to conclude with a practical scheme that has performance similar to that of the idealized setting.…”

“…As the read channel becomes more degraded, the threshold voltages are gradually increased to what would be the nominal values in a standard device not employing DVA. As shown in our precursor conference papers [22], [23], DVA maintains the required mutual information while significantly extending lifetime. Fig.…”

“…Based on [26]- [37], our precursor conference papers [22] and [23] propose a parameterized channel model characterizing the noise as having three additive components: programming noise, wear-out noise, and retention noise. This paper improves the model of [22] and [23] by adding the effects of cell-to-cell interference and programming error.…”

“…This paper begins by introducing DVA in an idealized setting and then removes ideal assumptions about channel knowledge and threshold resolution to conclude with a practical scheme that has performance similar to that of the idealized setting. DVA is first explored under the assumption of perfect knowledge of the channel state as in our precursor conference paper [22], except with a more complex read channel that includes cell-to-cell interference and programming errors.…”

Deep Learning-Aided Dynamic Read Thresholds Design for Multi-Level-Cell Flash Memories

“…They are caused by the leakage of the charges from the floating gate over time, which leads to a decrease of the threshold voltage. Following [20,28,29], the retention noise is assumed to follow the Gaussian distribution, given by…”