Regression Analysis for Optimal Read Voltage in Semiconductor Memory

Solution Overview

Problem

Semiconductor memory devices, particularly multi-level cell (MLC) flash memory devices, face challenges in accurately discriminating threshold voltage states due to overlapping voltage states, leading to difficulties in determining optimal read voltages for reliable data integrity without increasing read frequency.

Innovation Solution

A method involving a memory system that uses regression analysis to model a probability density function based on read operations with different voltages, determining the optimal read voltage by analyzing coordinate points and coefficients, thereby reducing the number of read operations required to decide on an optimal read level.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If multiple read operations are performed with different read voltages to determine optimal read level, then data integrity is improved, but read frequency increases and time consumption increases

Engineering Contradiction:

Improvedata integrityVSAvoidtime consumption

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent performs preliminary read operations with different read voltages to collect threshold voltage distribution data, then uses regression analysis to predict the optimal read voltage. This preliminary data collection and analysis allows subsequent reads to use the predetermined optimal voltage directly, avoiding repeated multiple read operations and reducing time consumption while maintaining data integrity.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical iterative process of repeatedly performing multiple read operations with a mathematical regression analysis system. By substituting the physical repeated reading process with a computational model that predicts optimal read voltage based on threshold voltage distribution, the system reduces time consumption while maintaining the reliability benefits of multiple voltage sampling.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If multiple read operations are performed with different read voltages to determine optimal read level, then data integrity is improved, but the number of read operations increases

Engineering Contradiction:

Improvedata integrityVSAvoidread frequency

Core Design Contradiction:

ReliabilityVSProductivity

Solution Approach 1:

The patent performs preliminary read operations with different read voltages to collect threshold voltage distribution data, then uses regression analysis to predict the optimal read voltage. This preliminary data collection and analysis allows subsequent reads to use the predetermined optimal voltage directly, avoiding repeated multiple read operations and reducing time consumption while maintaining data integrity.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical iterative process of repeatedly performing multiple read operations with a mathematical regression analysis system. By substituting the physical repeated reading process with a computational model that predicts optimal read voltage based on threshold voltage distribution, the system reduces time consumption while maintaining the reliability benefits of multiple voltage sampling.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If regression analysis is used to determine optimal read voltage, then the number of read operations is reduced, but computational complexity increases

Engineering Contradiction:

Improveread frequencyVSAvoidcomputational complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent extracts the computationally intensive regression analysis process from the real-time read operation and performs it as a separate preliminary step. By separating the computational complexity from the operational flow, the system can afford complex calculations during initialization while maintaining simple, fast read operations during actual data access, thus improving read frequency without permanently increasing operational complexity.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent performs preliminary read operations with different read voltages to collect threshold voltage distribution data, then uses regression analysis to predict the optimal read voltage. This preliminary data collection and analysis allows subsequent reads to use the predetermined optimal voltage directly, avoiding repeated multiple read operations and reducing time consumption while maintaining data integrity.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS9552887B2Semiconductor memory systems using regression analysis and read methods thereof

Publication Date: 2017.01.24 SAMSUNG ELECTRONICS CO LTD

AI summary

A memory system includes: a bit counter and a regression analyzer. The bit counter is configured to generate a plurality of count values based on data read from selected memory cells using a plurality of different read voltages, each of the plurality of count values being indicative of a number of memory cells of a memory device having threshold voltages between pairs of the plurality of different read voltages. The regression analyzer is configured to determine read voltage for the selected memory cells based on the plurality of count values using regression analysis.