Edge Device Control System for Machine Learning Workload Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing edge devices in networks are not effectively utilized for machine learning tasks due to inefficiencies in processing speed and resource allocation.
Innovation Solution
A computer system that detects and combines edge devices connected to a gateway, determines their processing performances, and allocates machine learning programs accordingly, enabling efficient execution of tasks based on device capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If edge devices are deployed in the network to perform machine learning, then processing speed is improved, but resource utilization is insufficient because some edge devices do not execute processing
Solution Approach 1:
The patent combines multiple edge devices into a collaborative machine learning system where devices are grouped based on their processing capabilities. By merging the computing resources of multiple edge devices and assigning them complementary roles in the machine learning pipeline, the system ensures that all devices contribute to processing tasks, thereby improving both processing speed and resource utilization simultaneously.
2Speed
If machine learning processing is distributed to edge devices, then processing speed improves, but task allocation efficiency deteriorates due to lack of coordination
Solution Approach 1:
The patent implements a feedback mechanism where the management device monitors the execution status and performance metrics of edge devices in real-time. Based on this feedback, the management device dynamically adjusts task allocation and redistributes workloads to optimize processing efficiency. This closed-loop control system simplifies task allocation by using automated performance-based decisions rather than complex manual coordination.
3Ease of operation
If all edge devices perform the same processing tasks, then ease of operation is improved, but processing speed deteriorates due to uniform resource utilization
Solution Approach 1:
The patent applies local quality by assigning different processing tasks to different edge devices based on their individual capabilities and characteristics. Instead of uniform task distribution, each edge device is optimized for specific types of processing operations (e.g., data collection, preprocessing, model training, inference), allowing the system to maintain operational simplicity while achieving high processing speeds through specialized resource allocation.
Data Source
AI summary
It is an object to provide a computer system, an edge device control method, and a program for improving a processing speed by effectively utilizing edge devices existing in a network. A computer system for controlling edge devices 100 connected to a gateway 200 to perform machine learning detects edge devices 100 connected to the gateway 200, determines a combination of the detected edge devices 100, acquires processing performances of the detected edge devices 100, determines programs for edge device 100 and a program for machine learning based on the determined combination and the acquired processing performances, transmits the determined programs for edge device 100 to the edge devices 100 in accordance with the acquired processing performances, causes the edge devices 100 to execute the programs for edge device 100 in accordance with the acquired processing performances, and causes a predetermined computer to execute the program for machine learning.