LDPC Decoding With Column-Weight LLR Updating for High HRER
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Irregular LDPC codes face challenges with high reliability error rates (HRER), where bits with errors are incorrectly deemed highly reliable, leading to incorrect decoding decisions, especially in scenarios with higher column weights, causing errors to spread across decoded data.
Innovation Solution
Employing two distinct sets of Log-Likelihood Ratios (LLRs) for decoding: LLR Set-1 for bits with lower column weights and LLR Set-2 for bits with higher column weights, adjusting confidence in bit values dynamically to improve error correction under high HRER conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single set of LLRs is used for all bits in irregular LDPC codes, then the decoding process is simple, but error correction performance deteriorates under high HRER conditions
Solution Approach 1:
The patent segments the LLR values into two distinct sets (first LLR set and second LLR set) based on column weight thresholds. Bits with column weights below the threshold use the first LLR set, while bits with column weights at or above the threshold use the second LLR set. This segmentation allows the decoder to handle different bit types with appropriate LLR values, improving error correction for high column weight bits without uniformly increasing complexity across all bits.
Solution Approach 2:
The patent applies different LLR sets to different local regions of the code based on column weight characteristics. High column weight bits (which are more susceptible to error propagation) receive specialized LLR treatment through the second LLR set, while other bits continue to use the first LLR set. This local quality approach targets error correction resources where they are most needed, improving overall reliability without uniformly increasing complexity.
2Reliability
If LLR values are adjusted for high column weight bits, then error correction improves under high HRER, but computational complexity increases
Solution Approach 1:
The patent divides the computational workload by segmenting LLR application based on column weight thresholds. Only bits with high column weights (at or above the threshold) receive the specialized second LLR set treatment, while the majority of bits with lower column weights continue to use the first LLR set. This segmentation reduces the overall computational energy required compared to applying complex LLR adjustments to all bits, while still targeting the specific bits that cause error propagation problems.
Solution Approach 2:
The patent changes the LLR parameter values specifically for high column weight bits by selecting from the second LLR set, which is designed to counteract the error propagation tendency of these bits. This parameter change is applied selectively based on column weight thresholds, improving error correction for critical bits while minimizing the overall computational energy increase by not applying the same changes to all bits.
3Measurement precision
If iterative parameter updating is performed for all bits, then decoding accuracy improves, but convergence time increases
Solution Approach 1:
The patent segments the iterative updating process by applying different LLR sets based on column weight thresholds. High column weight bits undergo iterative updating with the second LLR set to improve decoding accuracy for these error-prone bits, while other bits are handled with the first LLR set. This segmentation allows the decoder to focus computational effort on bits that require it, improving overall decoding accuracy without uniformly increasing convergence time across all bits.
Solution Approach 2:
The patent applies enhanced iterative parameter updating with specialized LLR sets locally to high column weight bits that exhibit error propagation characteristics. This local quality approach improves decoding accuracy specifically where needed (high column weight bits) while maintaining faster convergence for the majority of bits with lower column weights, thus optimizing the trade-off between accuracy and convergence time.
Data Source
AI summary
An error-handling module of a controller determines that a codeword received from a memory device has errors. The error-handing module assigs an initial set of log-likelihood ratios (LLRs) to bits of the codeword based on a column weight (CW) of a low density parity check (LDPC) code. The error-handling module updates the LLRs based on parity check outcomes. The controller decodes embedded data from the codeword based on the updated LLRs.