LDPC Soft Information Propagation for Multiple Word Line Failures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In memory systems, especially SSDs, soft information generated during LDPC decoding is not effectively utilized after initial error correction, leading to inefficiencies in error recovery, particularly when multiple word lines fail, as current methods do not adaptively adjust reliability based on iteration and unsatisfied check information.
Innovation Solution
Implementing a memory system with a memory controller that performs soft decoding across all word lines, generates extrinsic information for failed word lines, updates soft information based on this extrinsic information, and propagates it across these lines, using scaling factors that adjust based on unsatisfied check and iteration data to enhance LDPC decoding reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If soft decoding is performed on all word lines to generate codewords, then error correction capability is improved, but decoding time and computational complexity increase
Solution Approach 1:
The decoding process is segmented into multiple passes: an initial soft decoding pass on all word lines, followed by selective re-decoding only for failed word lines using propagated soft information from successfully decoded word lines. This segmentation reduces the overall decoding time while maintaining error correction capability.
Solution Approach 2:
Soft information is generated during the initial soft decoding pass for all word lines before the final decoding decision. This preliminary soft information is then propagated and reused for failed word lines, avoiding redundant computation and reducing total decoding time.
2Reliability
If extrinsic information is propagated across failed word lines with adaptive scaling, then decoding reliability is improved, but device complexity increases
Solution Approach 1:
The extrinsic information propagation uses dynamic scaling factors that are adaptively adjusted based on the number of unsatisfied checks and iteration count for each failed word line. This dynamic adjustment improves decoding reliability by weighting information according to its reliability without requiring complex hardware structures.
Solution Approach 2:
The scaling factor parameter is changed adaptively based on decoding performance metrics (unsatisfied checks and iteration count). This parameter adjustment allows the system to optimize decoding reliability for different failure scenarios without increasing fundamental device complexity.
3Reliability
If multiple decoding iterations are performed on failed word lines, then error recovery is improved, but energy consumption increases
Solution Approach 1:
Instead of performing multiple full decoding iterations on all word lines, the system performs partial re-decoding only on failed word lines using propagated soft information. This partial action achieves error recovery for failed lines while minimizing energy consumption by avoiding redundant processing of successfully decoded lines.
Solution Approach 2:
The decoding process uses feedback from the initial soft decoding results (soft information and unsatisfied check counts) to guide subsequent re-decoding decisions. This feedback mechanism allows the system to focus energy only on failed word lines that need additional decoding iterations, improving energy efficiency.
Data Source
AI summary
Systems, memory controllers, decoders and methods perform decoding by exploiting differences among word lines for which soft decoding fails (failed word lines). Such decoding generates extrinsic information for codewords of failed word lines based on the soft decoding. The soft information obtained during the soft decoding is updated based on the extrinsic information, and the updated soft information is propagated across failed word lines. Low-density parity-check (LDDC) decoding of codewords of failed word lines is performed with the updated soft information.