LDPC Decoder Asymmetric LLR Tracking for NAND Flash Errors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Non-volatile memory devices face challenges in maintaining data integrity due to noise introduced by program disturb and inter-cell interference, leading to voltage level drops over time, which affects the longevity and accuracy of memory cells, especially in NAND flash devices.
Innovation Solution
The implementation of a neural network-based method that estimates an asymmetric ratio using observables from a min-sum hard decoder, allowing for the adjustment of initial log-likelihood ratio (LLR) values and enabling an additional read retry, thereby optimizing read voltage thresholds to account for memory cell degradation and improve data retrieval across the lifespan of memory devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If symmetric LLRs are used in the message passing algorithm, then the decoder operation is simple and fast, but the data integrity deteriorates due to voltage drop and asymmetric errors in degraded memory cells
Solution Approach 1:
The patent applies dynamics by making the LLR initialization adaptive rather than static. The system dynamically switches between symmetric and asymmetric LLR initialization based on the degradation state of memory cells, detected through the asymmetric ratio metric. This allows the decoder to adapt its behavior to match the actual channel conditions, improving reliability without permanently sacrificing speed.
Solution Approach 2:
The patent changes the parameter of LLR initialization from fixed symmetric values to asymmetric values based on the asymmetric ratio. By modifying the initialization parameters according to the detected error pattern asymmetry, the system improves decoding accuracy for degraded memory cells while maintaining computational efficiency through conditional application.
2Reliability
If asymmetric LLRs are used to account for voltage drop and asymmetric errors, then data integrity improves, but the decoder complexity increases
Solution Approach 1:
The patent applies asymmetry by using asymmetric LLR initialization values that reflect the actual asymmetric error patterns in degraded memory cells. Instead of using symmetric initialization, the system employs different initial weights for bits affected by voltage drop, thereby improving reliability. The asymmetry is applied selectively based on the detected asymmetric ratio to manage complexity.
Solution Approach 2:
The patent applies partial action by using asymmetric LLR initialization only when needed, based on the asymmetric ratio threshold. Rather than always using complex asymmetric initialization, the system applies it partially only when the memory cells show signs of degradation, thereby improving reliability without incurring the full complexity cost in all operating conditions.
3Measurement precision
If the number of iterations is increased to improve decoding accuracy, then data retrieval accuracy improves, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by performing asymmetric ratio detection and LLR initialization adjustment before the main decoding iterations begin. This preliminary preparation optimizes the decoding process in advance, allowing fewer iterations to achieve the same accuracy, thereby reducing processing time while maintaining decoding accuracy.
Data Source
AI summary
Devices, systems, and methods for improving performance of an iterative decoder in a non-volatile memory are described. An example method includes receiving a noisy codeword that is based on a transmitted codeword generated from a low-density parity-check (LDPC) code, performing, on the noisy codeword for a first number of iterations, a message passing algorithm between a plurality of variable nodes and a plurality of check nodes that represent a parity matrix of the LDPC code, assigning, based on comparing an asymmetric ratio to a threshold, a set of asymmetric LLRs to a set of LLRs (that were initialized using symmetric LLRs) used in the message passing algorithm, and determining, subsequent to performing the message passing algorithm on the noisy codeword for a second number of iterations, a candidate version of the transmitted codeword.