Adaptive Check Node Approximation for LDPC Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LDPC decoding methods, such as the min-sum algorithm, suffer from performance degradation and require global attenuation factors, which are difficult to optimize, leading to suboptimal error correction in NAND-based flash memories.
Innovation Solution
An adaptive check node approximation method using the smallest and second smallest log-likelihood ratios (LLRs) to decode LDPC encoded data, eliminating the need for global attenuation factors and improving error correction capability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the sum-product algorithm (SPA) is used for LDPC decoding, then decoding performance is optimized, but computational complexity increases significantly
Solution Approach 1:
The patent transforms the sum-product algorithm into the min-sum algorithm by changing the parameter used in calculations from logarithmic domain addition to minimum magnitude selection. This parameter change simplifies the computational operations while maintaining the core decoding functionality, thereby reducing computational complexity while preserving acceptable decoding performance
Solution Approach 2:
The patent applies local quality by introducing position-dependent attenuation factors that are specifically tailored to different positions within the LDPC code structure. Instead of applying a uniform global attenuation factor, the system adjusts attenuation locally at each check node based on the specific characteristics of that position, thereby optimizing error correction performance without requiring the full computational overhead of the sum-product algorithm
2Device complexity
If the min-sum algorithm (MSA) is used to reduce computational complexity, then computational complexity decreases, but decoding performance degrades
Solution Approach 1:
The patent introduces attenuation factors as a parameter modification to the min-sum algorithm. By multiplying the min-sum results with position-specific attenuation factors, the system corrects the inherent performance degradation of the min-sum algorithm, bringing its results closer to the optimal sum-product algorithm performance while maintaining the computational simplicity of the min-sum approach
Solution Approach 2:
The patent implements local quality by applying different attenuation factors to different positions in the LDPC code. Each check node receives a customized attenuation factor based on its specific position and characteristics, allowing the system to optimize performance locally rather than applying a one-size-fits-all global attenuation factor, thereby improving overall decoding performance
3Reliability
If a global attenuation factor is applied to compensate for min-sum algorithm errors, then error correction improves, but optimization difficulty increases due to the indefinite nature of the factor
Solution Approach 1:
The patent resolves the optimization difficulty by transitioning from a global attenuation factor to position-specific local attenuation factors. Each attenuation factor is determined based on the specific characteristics of its corresponding check node position, eliminating the need for complex global optimization and making the system more adaptable to local error patterns without requiring iterative tuning of a single global parameter
Solution Approach 2:
The patent applies preliminary action by pre-determining the attenuation factors based on the LDPC code structure and position characteristics before the actual decoding process begins. This eliminates the need for real-time optimization during decoding, as the appropriate attenuation factors are already calculated and ready for application, thereby reducing optimization difficulty and improving processing speed
Data Source
AI summary
A low-density parity check (LDPC) decoder is provided for decoding low-density parity check (LDPC) encoded data wherein an adaptive check node approximation is performed at the check node processor utilizing the smallest magnitude log-likelihood ratio (LLR) and the second smallest magnitude log-likelihood ratio (LLR) to adapt to the current conditions at the check node.