Model-less Regression for Memory Read Threshold Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current memory systems face challenges in optimizing read threshold values, leading to suboptimal performance and endurance due to reliance on model-based schemes that assume specific distributions, which can be inflexible and less accurate.
Innovation Solution
A memory system and method utilizing model-less regression to estimate threshold voltage distribution curves based on measured probability values, determining optimal read threshold voltage values without assuming underlying distributions, thereby improving read threshold prediction accuracy and quality of service.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If model-based schemes are used to optimize read threshold values, then the optimization process has a structured approach, but the accuracy and adaptability deteriorate due to reliance on assumed distributions that may not match actual conditions
Solution Approach 1:
The patent changes the fundamental parameter of the optimization approach by transitioning from model-based methods (which assume specific statistical distributions) to model-less regression methods. This allows the system to adapt to actual threshold voltage distributions without being constrained by predefined models, thereby improving prediction accuracy while maintaining a systematic optimization process through regression analysis.
Solution Approach 2:
The patent implements a dynamic optimization approach where the read threshold values are continuously adjusted based on regression analysis of actual measured data. Instead of relying on static assumed distributions, the system dynamically adapts to changing conditions in the memory device, improving both accuracy and adaptability over time.
2Device complexity
If model-based schemes are used to optimize read threshold values, then the optimization process is simpler to implement, but the adaptability to varying conditions deteriorates
Solution Approach 1:
The patent changes the approach from fixed model-based parameters to adaptive model-less regression parameters. By using regression analysis on actual measured threshold voltage values, the system maintains relatively simple implementation while significantly improving adaptability to varying memory device conditions, program states, and wear levels.
Solution Approach 2:
The system performs self-adaptation by automatically adjusting read threshold values based on regression analysis of its own measured data. This self-service mechanism enables the memory device to optimize its own read operations without requiring external calibration or complex predefined models, balancing simplicity with adaptability.
3Speed
If pre-defined models are used for threshold optimization, then the processing speed is faster, but the prediction accuracy deteriorates due to distribution assumptions
Solution Approach 1:
The patent changes the optimization methodology from model-based to model-less regression, maintaining fast processing speeds through efficient regression calculations while dramatically improving threshold voltage distribution accuracy. The regression approach directly fits actual measured data without the computational overhead of complex statistical models.
Data Source
AI summary
A controller optimizes a read threshold value for a memory device using model-less regression. The controller performs read operations on cells using read threshold voltage values. The controller measures probability values for the multiple read threshold voltage values, and estimates a threshold voltage distribution curve based on the multiple read threshold voltage values and the measured probability values using a set regression formula. The controller determines a read threshold voltage value corresponding to a set point on the threshold voltage distribution curve, and performs a read operation on the cells using the read threshold voltage value.