In-NAND LLR Generation for Low-Latency Soft Read Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating log likelihood ratio (LLR) values in NAND storage systems involve multiple read operations, leading to high latency and data transfer times, which hinder efficient error correction performance.
Innovation Solution
The implementation of an In-NAND LLR generation scheme where a single soft read command triggers multiple reads within the NAND storage medium to generate LLR values internally, using look-up tables to determine LLR values for each bin, reducing the need for external data transfer and minimizing latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple read operations are performed externally to generate LLR values, then error correction performance is improved, but latency and data transfer time increase
Solution Approach 1:
The patent combines multiple read operations and LLR value generation into a single integrated process performed within the NAND storage medium itself. The processor inside the storage medium executes multiple internal reads and generates all necessary LLR values in one operation, eliminating the need for separate external read commands and reducing total latency while maintaining error correction performance.
Solution Approach 2:
The patent introduces an intermediary processor within the NAND storage medium that acts as a bridge between the storage cells and the external controller. This internal processor performs the complex task of executing multiple reads and generating LLR values, shielding the external controller from the complexity and time requirements of multiple read operations.
2Measurement precision
If multiple read operations are performed to generate LLR values, then accuracy of LLR values is improved, but data transfer amount increases
Solution Approach 1:
The patent extracts the LLR value generation process from the external controller and relocates it entirely within the NAND storage medium. The internal processor performs all necessary read operations and computations internally, extracting only the final LLR values for external transfer. This eliminates the need to transfer multiple sets of raw read data while maintaining full accuracy of the generated LLR values.
3Device complexity
If LLR generation is performed externally, then system complexity is reduced, but processing speed decreases
Solution Approach 1:
The patent implements self-service by enabling the NAND storage medium to autonomously perform LLR value generation through its internal processor. The storage medium serves itself by executing multiple internal reads and computing LLR values without external intervention for each read operation, significantly improving processing speed while the external controller maintains simplified oversight functionality.
Data Source
AI summary
A memory system and a method for generation of log likelihood ratio LLR values. The system has a storage medium having therein a processor, and a memory controller of the memory system configured to send a soft read command to the storage medium. The processor in the storage medium, in response to the soft read command, is configured to read pages of data stored in the storage medium and generate within the storage medium the LLR values.