MLC Flash Memory Read Reliability via Soft Decision Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multi-level cell flash memory devices face reliability issues and increased read failure rates due to charge loss and interference between memory cells, making it difficult to accurately read data stored in overlapping threshold voltage distributions.
Innovation Solution
A method is introduced that involves reading hard decision data and soft decision data, where the soft decision data indicates the reliability of the hard decision data, and using a de-randomizer to combine these with random sequences to improve data accuracy. The soft decision data bypasses the de-randomizer during certain operations to provide reliability information for error correction, enhancing read resolution and reducing errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If multi-level cell flash memory devices store more bits per cell to increase storage capacity, then integration density is improved, but reliability decreases and read failure rate increases
Solution Approach 1:
The patent segments the data reading process into two independent paths: hard decision data reading and soft decision data reading. The soft decision data path separately captures reliability information without being affected by the randomization applied to hard decision data, enabling independent optimization of both storage capacity and read reliability.
Solution Approach 2:
The patent introduces soft decision data as an intermediary that carries reliability information about the stored data states. This intermediary allows the system to distinguish between confidently stored data and uncertain data, enabling selective error correction without compromising the high-density storage capability of MLC cells.
2Reliability
If randomization is applied to hard decision data to improve read reliability, then error correction capability is improved, but soft decision data accuracy deteriorates if de-randomized
Solution Approach 1:
The patent segments the data processing paths so that hard decision data undergoes randomization for error correction, while soft decision data bypasses the randomization process. This segmentation allows each data type to be processed according to its specific requirements, maintaining both error correction capability and reliability information accuracy.
Solution Approach 2:
The patent extracts the reliability information into a separate soft decision data path that is independent from the randomization process. By taking out the reliability information extraction from the hard decision data path, the system preserves the integrity of reliability indicators while still benefiting from randomization-based error correction.
3Device complexity
If de-randomizer processes both hard decision data and soft decision data, then processing simplicity is improved, but data integrity deteriorates due to incorrect random sequence application
Solution Approach 1:
The patent segments the de-randomizer functionality to apply random sequence processing only to hard decision data while allowing soft decision data to bypass this processing stage. This segmentation prevents incorrect random sequence application to reliability information, maintaining data integrity while keeping the overall processing architecture relatively simple.
Data Source
AI summary
An apparatus and a method for reading from a non-volatile memory whereby soft decision data is used to determine the reliability of hard decision data. The hard decision data read from the non-volatile memory is de-randomized and the soft decision data read from the non-volatile memory is not de-randomized. Using the soft decision data, the hard decision data is decoded.