Integrated Sensor Device with Deep Learning Accelerator and RAM

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing sensor devices face challenges in efficiently processing artificial neural networks (ANNs) due to high energy consumption and computation time, limiting their performance in applications requiring real-time data processing and inference.

Innovation Solution

Integration of a deep learning accelerator (DLA) with random access memory in sensor devices, which includes specialized hardware for parallel vector and matrix calculations, reducing the need for central processing unit assistance and optimizing data access through high communication bandwidth.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If sensor devices use general-purpose processors for ANN computations, then device complexity is reduced, but energy consumption increases and computation time increases

Engineering Contradiction:

Improvedevice complexityVSAvoidenergy consumption

Core Design Contradiction:

Device complexityVSUse of energy by moving object

Solution Approach 1:

The system is segmented into distinct functional units: a dedicated deep learning accelerator for ANN computations, random access memory for data storage, and a host system for overall control. This segmentation allows the computationally intensive ANN operations to be handled by specialized hardware while keeping the overall device architecture manageable.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

A deep learning accelerator acts as an intermediary component between the sensor device and the host system. This accelerator handles the computationally intensive ANN computations locally, reducing the need for continuous host system intervention and lowering overall energy consumption while maintaining independence.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Device complexity

If sensor devices use general-purpose processors for ANN computations, then device complexity is reduced, but computation time increases

Engineering Contradiction:

Improvedevice complexityVSAvoidcomputation time

Core Design Contradiction:

Device complexityVSLoss of time

Solution Approach 1:

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent replaces general-purpose mechanical processing with specialized electronic hardware acceleration. The deep learning accelerator uses dedicated electronic circuits optimized for matrix and vector operations, significantly speeding up ANN computations compared to general-purpose processors.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Use of energy by moving object

If sensor devices integrate deep learning accelerator and random access memory, then energy consumption is reduced and computation time is reduced, but device complexity increases

Engineering Contradiction:

Improveenergy consumptionVSAvoiddevice complexity

Core Design Contradiction:

Use of energy by moving objectVSDevice complexity

Solution Approach 1:

The deep learning accelerator and random access memory are merged into a single integrated device, allowing them to work together as a unified system. This integration reduces communication overhead and enables efficient data exchange between the accelerator and memory, optimizing performance while managing complexity through unified architecture.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The integrated device provides self-service capabilities by performing ANN computations locally using the deep learning accelerator and stored data from random access memory. This eliminates the need for continuous external processing, allowing the device to independently handle computationally intensive tasks and reduce overall energy consumption.

Inventive Principle:

Principle #25Self-service

4Loss of time

If sensor devices integrate deep learning accelerator and random access memory, then computation time is reduced, but device complexity increases

Engineering Contradiction:

Improvecomputation timeVSAvoiddevice complexity

Core Design Contradiction:

Loss of timeVSDevice complexity

Solution Approach 1:

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS20230161936A1Integrated Sensor Device with Deep Learning Accelerator and Random Access Memory

Publication Date: 2023.05.25 MICRON TECHNOLOGY INC

AI summary

Systems, devices, and methods related to a deep learning accelerator and memory are described. For example, an integrated sensor device may be configured to execute instructions with matrix operands and configured with: a sensor to generate measurements of stimuli; random access memory to store instructions executable by the deep learning accelerator and store matrices of an artificial neural network; a host interface connectable to a host system; and a controller to store the measurements generated by the sensor into the random access memory as an input to the artificial neural network. After the deep learning accelerator generates in the random access memory an output of the artificial neural network by executing the instructions to process the input, the controller may communicate the output to a host system through the host interface.