On-Device AI Processing Load Distribution for Speed and Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

On-device AI processing faces limitations in processing speed and accuracy, particularly in noisy environments or with long texts, compared to server or cloud-based processing.

Innovation Solution

An on-device AI processing system that distributes AI processing load to external or internal devices when the terminal's capacity is exceeded, combining local and distributed processing results to enhance performance.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If on-device AI processing is used, then processing speed is improved, but processing accuracy deteriorates

Engineering Contradiction:
Improveprocessing speedVSAvoidprocessing accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The patent segments AI processing into two parts: simple tasks are processed locally on the terminal device for speed, while complex tasks are processed externally for accuracy. This segmentation allows the system to optimize for both speed and accuracy depending on the task requirements.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary mechanism that determines whether to process AI tasks locally or externally based on task complexity. This intermediary selection process enables the system to achieve both fast processing for simple tasks and high accuracy for complex tasks.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Productivity

If on-device AI processing is used, then processing capability is improved, but power consumption increases

Engineering Contradiction:
Improveprocessing capabilityVSAvoidpower consumption
Core Design Contradiction:
ProductivityVSUse of energy by moving object

Solution Approach 1:

The patent segments processing tasks based on their computational requirements, assigning simple tasks to local processing and complex tasks to external processing. This segmentation enables the terminal to maintain processing capability while avoiding excessive power consumption for tasks that don't require full local processing power.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial processing locally and external processing for the remainder, rather than always using full local processing capability. This partial action approach maintains necessary processing capability while reducing overall power consumption by avoiding excessive local computation.

Inventive Principle:
Principle #16Partial or excessive action

3Loss of time

If on-device AI processing is used, then response time is improved, but processing completeness deteriorates

Engineering Contradiction:
Improveresponse timeVSAvoidprocessing completeness
Core Design Contradiction:
Loss of timeVSLoss of information

Solution Approach 1:

The patent segments processing based on task complexity, handling simple tasks locally for fast response and complex tasks externally for complete processing. This segmentation ensures that response time is optimized for simple tasks while processing completeness is maintained for complex tasks.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary selection mechanism that determines whether to process tasks locally or externally based on complexity assessment. This intermediary process ensures that simple tasks get fast local response while complex tasks receive complete external processing.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4597319A1On-device artificial intelligence processing distributed processing device and method
Publication Date: 2025.08.06 LX SEMICON CO LTD
  • EP4597319A1 patent drawingFigure 1
  • EP4597319A1 patent drawingFigure 2
  • EP4597319A1 patent drawingFigure 3

AI summary

An on-device artificial intelligence (AI) processing distributed processing device and method capable of distributing AI processing load of a terminal through internal or external other devices are provided. The on-device AI processing distributed processing device can, measure an amount of AI processing for performing the user command, select an AI processing distributed processing target when the measured amount of AI processing exceeds a self-processing capacity, request distributed process for AI processing to the selected AI processing distributed processing target, and when a first AI processing result value is received from the AI processing distributed processing target, provide a final result value based on the first AI processing result value and a self-processed second AI processing result value.