SSD Controller LLR Lookup Tables for NAND Soft Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional solid-state drive controllers face challenges in maintaining reliable error correction due to poor raw bit error rates in NAND flash memory, as they lack natural soft decision values necessary for low density parity check codes used in soft decoding.
Innovation Solution
An apparatus with lookup tables storing log likelihood ratio values is implemented in a controller, allowing for soft decoding of codewords by converting hard decisions from flash devices into decoding parameters, which can operate independently of flash cell voltage distributions and channel properties, and supports different numbers of retry reads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional error correction codes are used without soft decoding capability, then device complexity is reduced, but error correction reliability deteriorates due to poor raw bit error rates in NAND flash memory
Solution Approach 1:
The patent introduces log likelihood ratio (LLR) values as intermediary soft decision parameters that bridge the gap between hard read data and soft decoding requirements. These LLR values are computed from read data and channel information, then fed to the decoder to enable soft decoding capability, thereby improving error correction reliability without requiring direct access to complex voltage distribution information
Solution Approach 2:
The patent transforms the decoding approach by changing from hard decision decoding to soft decision decoding. This parameter change involves using probability-based LLR values instead of binary decisions, allowing the decoder to utilize magnitude information from read operations to improve error correction performance while maintaining manageable complexity through standardized LLR computation methods
2Reliability
If soft decoding with log likelihood ratio values is implemented, then error correction capability is improved, but device complexity increases due to the need for lookup tables and additional processing
Solution Approach 1:
The patent pre-computes and stores log likelihood ratio values in lookup tables during system initialization or offline processing. This preliminary action allows the actual decoding operation to simply retrieve pre-computed values rather than performing complex real-time calculations, thereby reducing runtime complexity while maintaining soft decoding capability
Solution Approach 2:
The patent uses lookup tables that store copies of pre-computed log likelihood ratio values for different read scenarios. Instead of recalculating complex probability distributions during each read operation, the system retrieves copied LLR values from memory, significantly reducing processing complexity while preserving the benefits of soft decoding
3Reliability
If multiple retry reads are performed to improve decoding accuracy, then error correction reliability is improved, but loss of time increases due to additional read operations
Solution Approach 1:
The patent implements dynamic adaptation of the decoding process based on the number of retry reads performed. The lookup tables are structured to provide different LLR values corresponding to different read scenarios (single read, multiple reads). This dynamic approach allows the system to optimize between time and accuracy by selecting appropriate LLR values based on actual read history, rather than always performing maximum retry reads
Data Source
AI summary
An apparatus having one or more lookup tables and a decoder is disclosed. The lookup tables are configured to store a plurality of sets of values of log likelihood ratios. The decoder is configured to (i) receive a codeword read from a memory, (ii) receive an initial one of the sets from the lookup tables and (iii) generate read data by decoding the codeword based on the values.