LDPC Soft Decoding Tuned to Memory Failure Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing soft bit decoding schemes for memory devices do not account for the dynamic nature of failure modes, leading to suboptimal performance and increased processing burden and latency.
Innovation Solution
A machine learning model is trained to identify optimal soft bit read parameters for LDPC decoding based on hard bit read positions derived from memory devices under different failure modes, allowing for dynamic adaptation of decoding parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fixed values are used for soft bit read positions and log likelihood ratios in soft bit decoding, then processing burden and latency are reduced, but soft decoding performance degrades due to inability to account for dynamic failure modes
Solution Approach 1:
The patent implements dynamic selection of soft bit read positions and log likelihood ratio values based on detected failure modes. Instead of using fixed values, the system adapts parameters in real-time according to the specific failure mode identified through hard bit decoding analysis, thereby optimizing decoding performance for each operational condition.
Solution Approach 2:
The system changes key decoding parameters (soft bit read positions and log likelihood ratio values) based on the detected failure mode. Multiple sets of parameters are pre-computed for different failure modes, and the appropriate set is selected dynamically, allowing the decoder to optimize performance for each specific failure scenario.
2Productivity
If fixed values are used for soft bit read positions and log likelihood ratios, then device complexity is reduced, but latency increases due to suboptimal performance requiring additional processing
Solution Approach 1:
Multiple sets of soft bit read positions and log likelihood ratio values are pre-computed and stored for different failure modes before actual decoding operations. This preliminary preparation allows the system to quickly select the appropriate parameter set based on detected failure modes without performing complex real-time calculations, thereby reducing latency while maintaining high throughput.
3Reliability
If dynamic adaptation of decoding parameters is implemented, then error correction performance is improved, but processing complexity increases
Solution Approach 1:
The patent segments the decoding process into distinct stages: hard bit decoding to identify failure modes, failure mode classification, and selective application of pre-computed parameter sets. This segmentation allows dynamic adaptation without requiring complex real-time optimization, as each stage handles a specific aspect of the problem independently.
Solution Approach 2:
Instead of performing complex real-time optimization, the system creates multiple copies of parameter sets (soft bit read positions and log likelihood ratios) for different failure modes. The appropriate copy is selected based on detected conditions, avoiding the need for complex computational processes while achieving adaptive optimization.
Data Source
AI summary
In some implementations, a device may receive a data signal from a memory device. The device may perform a low-density parity check (LDPC) hard bit decoding on the data signal to identify a plurality of hard bit read positions (HBRPs). The device may identify, with a machine learning model using the plurality of HBRPs, a failure mode of the memory device. The device may identify a set of parameters for an LDPC soft bit decoding based on the failure mode. The device may perform the LDPC soft bit decoding on the data signal using the set of parameters.