Flash Memory Gaussian Sampling Using Threshold Voltage Deviation

Solution Overview

Problem

Existing methods for generating Gaussian error data for lattice-based cryptography, such as those using true random number generators and FPGA or software circuits, incur significant overhead in terms of implementation size and speed, making them inefficient for widespread adoption.

Innovation Solution

A method and apparatus utilizing flash memory to generate Gaussian error data by leveraging the threshold voltage deviation, which is inherent in the physical characteristics of semiconductor devices, thereby eliminating the need for true random number generators, FPGA circuits, or software circuits, and reducing production costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If true random number generator, FPGA circuit, or software circuit is used to generate Gaussian error data, then Gaussian error data can be generated for lattice-based cryptography, but implementation size and speed overhead increases significantly

Engineering Contradiction:

ImproveGaussian error data generation capabilityVSAvoidimplementation size

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The flash memory device generates Gaussian error data using its own inherent threshold voltage deviation characteristics during normal read operations, without requiring external random number generators or additional circuitry. The device serves itself by utilizing its natural physical variations to produce the required error data for lattice-based cryptography

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent extracts the Gaussian error data generation function from separate dedicated components (true random number generators, FPGA circuits) and integrates it into the flash memory device itself by utilizing the threshold voltage deviation that naturally occurs during read operations

Inventive Principle:

Principle #2Taking out (Extraction)

2Reliability

If true random number generator, FPGA circuit, or software circuit is used to generate Gaussian error data, then Gaussian error data can be generated for lattice-based cryptography, but speed overhead increases

Engineering Contradiction:

ImproveGaussian error data generation capabilityVSAvoidread/write speed

Core Design Contradiction:

ReliabilityVSSpeed

Solution Approach 1:

The patent merges two functions into a single process: the normal read operation of flash memory and the generation of Gaussian error data. By combining these functions, the system achieves both memory access and error data generation simultaneously, eliminating the speed penalty of separate generation processes

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The flash memory device generates Gaussian error data using its own inherent threshold voltage deviation characteristics during normal read operations, without requiring external random number generators or additional circuitry. The device serves itself by utilizing its natural physical variations to produce the required error data for lattice-based cryptography

Inventive Principle:

Principle #25Self-service

3Reliability

If dedicated Gaussian extractor components are used, then Gaussian error data generation is reliable, but production cost increases

Engineering Contradiction:

ImproveGaussian error data generation capabilityVSAvoidproduction cost

Core Design Contradiction:

ReliabilityVSEase of manufacture

Solution Approach 1:

The flash memory device performs multiple functions: it serves as both the storage medium and the Gaussian error data generator. This multi-functionality eliminates the need for separate dedicated components, thereby reducing production costs while maintaining reliability

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The flash memory device generates Gaussian error data using its own inherent threshold voltage deviation characteristics during normal read operations, without requiring external random number generators or additional circuitry. The device serves itself by utilizing its natural physical variations to produce the required error data for lattice-based cryptography

Inventive Principle:

Principle #25Self-service

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables the implementation of a high-integration, fast-read/write Gaussian sampler with reduced overhead, utilizing commercially available flash memory to generate Gaussian-distributed error signals essential for lattice-based cryptography, thus addressing the inefficiencies of previous methods.

Implementation Method 1

uses a physical characteristic in which the threshold voltage of the transistor varies depending on whether a charge is present in the charge-trapping layer

Methodology Applied

Scientific EffectThreshold voltage deviation:

Data Source

PatentUS11880269B2Method for generating Gaussian error data using flash memory and apparatus using the same

Publication Date: 2024.01.23 ELECTRONICS & TELECOMM RES INST

AI summary

Disclosed herein are a method for generating Gaussian error data using flash memory and an apparatus using the method. The method includes receiving a request to generate Gaussian error data and delivering an operation command to flash memory; generating Gaussian error noise based on a threshold voltage that is generated when the flash memory performs the operation command; and generating Gaussian error data so as to correspond to the Gaussian error noise and providing the same.