Memory Device Lifetime Estimation Without Usage History
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor memory devices face challenges in accurately determining their remaining useful lifetime due to difficulty in recognizing usage history, which is crucial for recertification and reuse processes.
Innovation Solution
A modeling method that estimates the used lifetime of memory devices without a usage history, using an accelerated aging test on unused memory devices to gather performance data, apply statistical distribution approximation, and train a lifetime calculation model based on conditional probabilities, without additional on-chip IP or dedicated ICs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If accelerated aging test is performed on unused memory devices to gather performance data, then the lifetime estimation accuracy is improved, but the test time and resource consumption increase
Solution Approach 1:
The patent performs accelerated aging tests on unused memory devices before they are deployed to gather performance degradation data. This preliminary action enables the construction of lifetime estimation models that can predict the remaining useful lifetime of memory devices without requiring actual usage history, thereby improving estimation accuracy while managing test time through proactive data collection
Solution Approach 2:
The patent applies accelerated aging conditions (such as elevated temperature, voltage, or stress levels) during testing to accelerate the aging process. By changing the operating parameters to more severe conditions, the patent can gather sufficient lifetime data in a shorter time frame, resolving the contradiction between measurement precision and time loss
2Measurement precision
If statistical distribution approximation is applied to performance measurement data, then the lifetime calculation model accuracy is improved, but the computational complexity increases
Solution Approach 1:
The patent transforms the complex performance measurement data into simplified statistical parameters (mean, standard deviation, skewness, kurtosis) through distribution approximation. This parameter transformation reduces the computational complexity while maintaining the essential characteristics of the data needed for accurate lifetime estimation
Solution Approach 2:
The patent extracts key statistical features from the complex performance measurement data and uses only these extracted features for model training. By taking out only the essential statistical parameters rather than processing the entire raw data set, the patent reduces computational complexity while preserving measurement precision
3Reliability
If performance data from multiple memory devices is collected and processed, then the estimation model reliability is improved, but the data processing time and resources increase
Solution Approach 1:
The patent segments the data processing into distinct stages: data collection during accelerated aging tests, statistical parameter extraction, model training, and validation. This segmentation allows for efficient processing by handling data in manageable chunks and using parallel processing for multiple memory devices, thereby improving model reliability while controlling data processing time
Solution Approach 2:
The patent uses multiple unused memory devices as replicates to gather statistical data. By collecting data from multiple identical units undergoing accelerated aging, the patent builds a reliable estimation model through statistical aggregation. This copying approach allows parallel data collection and processing, improving reliability without linearly increasing processing time
Data Source
AI summary
In an example modeling method for estimating used lifetime of a memory device, a plurality of performance measurement data associated with a plurality of performances of a plurality of unused memory devices are obtained based on an accelerated aging test performing on the plurality of unused memory devices. A plurality of statistical data are calculated based on performing a statistical distribution approximation on the plurality of performance measurement data. A plurality of conditional probabilities are calculated based on a plurality of sample performance data associated with the plurality of performances and the plurality of statistical data. A lifetime calculation model is trained based on the plurality of conditional probabilities. The lifetime calculation model outputs estimated used lifetime data and uncertainty data. The estimated used lifetime data corresponds to the plurality of sample performance data. The uncertainty data represents uncertainty of the estimated used lifetime data.