Optimal Read Threshold Estimation Using Distribution Symmetry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for estimating the optimal read threshold in solid state storage systems assume symmetry in conditional distribution functions, which breaks down over time due to wear, and often fail to find a plausible minimum, leading to inaccurate read thresholds.
Innovation Solution
A method that estimates the optimal read threshold by using symmetry in conditional distribution functions, specifically by finding an axis of symmetry in sampled distributions and generating mirrored copies to determine the optimal read threshold without relying on finding a minimum, suitable for both single-level cell and multi-level cell storage systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If minimum-finding techniques are used to estimate optimal read threshold, then the estimation process is simple, but the accuracy deteriorates when conditional distribution functions are not symmetric
Solution Approach 1:
The patent applies asymmetry by recognizing that conditional distribution functions in worn storage systems are not symmetric, and therefore develops an estimation technique that does not assume symmetry. The method uses the relationship between cumulative distribution functions and probability density functions to accurately determine the optimal read threshold even when the underlying distributions are asymmetric, thus resolving the contradiction between simple estimation and accurate results.
2Adaptability or versatility
If traditional minimum-finding methods are used, then the approach works for fresh storage, but it fails when storage is worn and distributions become asymmetric
Solution Approach 1:
The patent applies dynamics by creating an adaptive estimation method that adjusts to the actual state of the storage system. Rather than relying on fixed assumptions about symmetry that work for fresh storage, the dynamic method uses measured cumulative distribution data to empirically determine the optimal read threshold, allowing it to adapt to worn storage conditions where distributions become asymmetric.
3Ease of operation
If symmetry assumption is made in distribution functions, then estimation is straightforward, but the result becomes inaccurate as storage ages and insulation breaks down
Solution Approach 1:
The patent applies self-service by enabling the system to determine its own optimal read threshold based on actual measured data from the storage cells. Instead of relying on pre-programmed symmetry assumptions, the method uses the storage system's own cumulative distribution measurements to empirically identify the threshold that minimizes bit errors, allowing the system to self-adjust as it ages.
4Measurement precision
If multiple test reads are performed to find a minimum, then more data is collected, but the process becomes slower and more complex
Solution Approach 1:
The patent applies taking out by extracting only the essential information needed for threshold estimation from the multiple test reads. Rather than performing complex minimization algorithms on all collected data, the method directly uses the cumulative distribution function values at the tested thresholds to determine the optimal threshold through a simpler relationship, reducing processing requirements while maintaining accuracy.
Data Source
AI summary
A group of cells in solid state storage is sampled in order to obtain a sampled distribution associated with the group of cells. An axis of symmetry is determined using the sampled distribution. At least some portion of a conditional distribution is determined based at least in part on: (1) the sampled distribution and (2) the axis of symmetry. An optimal read threshold is estimated using the sampled distribution and the conditional distribution.