NAND Stripe Read Recovery Using XOR-Derived LDPC Soft Bits

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing read recovery methods in memory sub-systems, particularly in non-volatile memory devices like NAND, face challenges in efficiently recovering failed translation units (TUs) without incurring the resource-intensive RAIN recovery operation.

Innovation Solution

The implementation of an improved decoder capability using the 1H3S LDPC decoding algorithm, which generates a third soft bit by performing an XOR operation between a RAIN drop and a raw read of a failed TU, thereby enhancing the likelihood of recovering the failed TU without necessitating a RAIN recovery operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If RAIN recovery operation is performed to recover failed translation units, then data recovery reliability is improved, but resource consumption increases and latency increases

Engineering Contradiction:

Improvedata recovery reliabilityVSAvoidresource consumption

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies partial action by implementing an enhanced decoder that attempts to recover failed translation units using improved LDPC decoding algorithms before resorting to the resource-intensive RAIN recovery operation. The decoder tries multiple decoding attempts with different parameters, performing only the necessary portion of the full RAIN recovery process to achieve successful data recovery in many cases without triggering the complete RAIN operation.

Inventive Principle:

Principle #16Partial or excessive action

2Reliability

If RAIN recovery operation is performed to recover failed translation units, then data recovery reliability is improved, but operation latency increases

Engineering Contradiction:

Improvedata recovery reliabilityVSAvoidoperation latency

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-calculating and storing syndrome values and implementing an enhanced decoder with optimized algorithms before actual data recovery is needed. When a translation unit fails, the pre-prepared decoder structure and pre-computed syndromes enable faster recovery attempts, reducing the time required before initiating RAIN recovery or achieving successful recovery.

Inventive Principle:

Principle #10Preliminary action

3Reliability

If enhanced decoder capability with 1H3S LDPC decoding is implemented, then recovery success rate improves, but device complexity increases

Engineering Contradiction:

Improverecovery success rateVSAvoiddecoder complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the complex 1H3S LDPC decoding process into distinct modular components: syndrome calculation module, iterative decoding module, and fallback to RAIN recovery module. Each component handles a specific aspect of the decoding process, making the overall complex system more manageable and maintainable while achieving high recovery success rates through the coordinated operation of these segmented functional blocks.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS12298852B2Read recovery for redundant array of independent NAND stripes

Publication Date: 2025.05.13 MICRON TECHNOLOGY INC

AI summary

A sign bit of a low-density parity-check (LDPC) codeword associated with a translation unit (TU) can be generated by performing an XOR operation on a RAIN drop corresponding to the TU and a raw read of the TU. The LDPC codeword can include a hard bit and three soft bits that include the sign bit. The LDPC codeword can be decoded using the hard bit and the three soft bits. A read recovery operation can be performed on the TU using the decoded LDPC codeword.