Hard-Decision LDPC Decoding With Dynamic Voting for Low BER
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional hard-decision Low Density Parity Check (LDPC) decoders struggle to correct errors effectively in low Bit Error Rate (BER) situations, particularly in NAND flash memories, where soft-decision decoding may not provide sufficient improvement and can lead to page read errors even with low error rates, due to the imprecision in charge storage and variations in cell parameters.
Innovation Solution
A hard-decision LDPC decoder with a dynamically adjustable voting method that strengthens or weakens the bit flipping requirements based on the degree of variable nodes and previous syndrome values, improving decoding capabilities in low BER regimes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional hard-decision LDPC decoding is used, then device complexity is low and ease of operation is good, but error correction capability deteriorates in low BER situations
Solution Approach 1:
The patent implements dynamic adjustment of the voting threshold parameter during the decoding process. The threshold is not fixed but adapts based on the current decoding state and syndrome information, allowing the decoder to optimize its error correction capability for low BER situations while maintaining hard-decision simplicity. This dynamic parameter adjustment resolves the contradiction by enabling improved reliability without increasing fundamental device complexity.
Solution Approach 2:
The invention changes the voting threshold parameter dynamically during decoding iterations. By adjusting this parameter based on syndrome weight and iteration count, the decoder can adapt to low BER conditions and improve error correction capability. This parameter change approach maintains the simplicity of hard-decision decoding while achieving better reliability performance.
2Reliability
If soft-decision decoding is used to improve error correction, then reliability improves, but power consumption and latency increase
Solution Approach 1:
The patent uses a hard-decision decoder with dynamically adjusted voting thresholds as a 'cheap' alternative to expensive soft-decision decoders. By optimizing the voting threshold parameter, the hard-decision approach achieves improved error correction capability without incurring the high power consumption and latency costs associated with soft-decision decoding. This resolves the contradiction by providing a low-cost solution that delivers sufficient reliability improvement.
3Reliability
If the voting threshold is strengthened to reduce false corrections, then reliability improves, but the number of decoding iterations increases leading to higher latency
Solution Approach 1:
The patent implements periodic adjustment of the voting threshold parameter during decoding iterations. The threshold is adjusted at specific intervals or based on syndrome weight thresholds, creating a rhythmic pattern of threshold changes that balances correction accuracy with iteration efficiency. This periodic action resolves the contradiction by preventing both excessive false corrections and unnecessary iteration delays.
Solution Approach 2:
The invention uses feedback from syndrome weight and decoding iteration count to dynamically adjust the voting threshold. This feedback mechanism allows the decoder to adapt the threshold in real-time, strengthening it when needed to prevent false corrections and weakening it when appropriate to maintain fast convergence. This feedback-controlled approach resolves the contradiction between decoding accuracy and latency.
Data Source
AI summary
A non-volatile memory controller includes a hard-decision Low Density Parity Check (LDPC) decoder with a capability to dynamically select a voting method to improve the decoding in low bit error rate (BER) situations. The hard-decision LDPC decoder dynamically selects a voting method associated with a strength requirement for bit flipping decisions. In one implementation, the voting method is selected based on the degree of a variable node and previous syndrome values.