Edge Computing Task Scheduling via Historical State Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dynamically scheduling computing tasks in mobile networks require significant training data and computing resources, making them inefficient for edge computing environments with limited resources.
Innovation Solution
A method that compares a current computing device's state to historical states to identify a matching historical scheduling policy, allowing for the allocation of computing tasks to other devices without the need for extensive machine learning computations, thereby reducing computational load and improving task scheduling efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If machine learning-based dynamic scheduling methods are used to allocate computing tasks in mobile networks, then task scheduling accuracy and adaptability are improved, but computing resource consumption and training data requirements increase significantly
Solution Approach 1:
The system pre-collects and stores historical state data and scheduling policy data in advance. When task scheduling is needed, it directly queries and applies pre-stored policies based on current state matching, avoiding real-time machine learning computations and reducing energy consumption while maintaining scheduling quality.
Solution Approach 2:
The system creates simplified copies of historical states and scheduling policies, storing them in a database for quick retrieval. Instead of performing complex machine learning inference on current states, it matches current states against historical state copies and applies corresponding policy copies, significantly reducing computing resource requirements.
2Productivity
If machine learning models are trained to optimize task scheduling, then scheduling performance is improved, but training time and computational overhead increase
Solution Approach 1:
Scheduling policies are determined and stored in advance based on historical data, eliminating the need for real-time model training. The system performs policy determination offline and stores results for immediate application, achieving high scheduling efficiency without training time delays.
Solution Approach 2:
The system extracts essential scheduling patterns from historical data and stores them as discrete policies in a database. This extraction approach separates the learning phase (offline policy determination) from the execution phase (online policy application), eliminating training time during actual task scheduling operations.
3Measurement precision
If complex scheduling algorithms are deployed at edge computing devices, then task allocation accuracy is improved, but device complexity and resource requirements increase
Solution Approach 1:
The system introduces a scheduling policy database as an intermediary between state detection and task allocation. Instead of implementing complex algorithms directly at edge devices, the database provides pre-determined policies that simplify the decision-making process at edge devices while maintaining allocation accuracy through carefully curated policy data.
Solution Approach 2:
The system stores simplified copies of scheduling policies in a database that can be easily queried and applied. These policy copies contain essential allocation rules without the computational complexity of full machine learning models, enabling accurate task allocation at resource-constrained edge devices.
Data Source
AI summary
The present disclosure relates to a method, a device, and a program product for managing a computing system. In a method, a current state and a plurality of historical states of a computing device in a computing system are acquired, the plurality of historical states respectively describing historical states of the computing device in the computing system at a plurality of historical time points. In response to determining that the current state matches a scheduling type for scheduling the computing device, the plurality of historical states are searched for a historical state matching the current state. A historical scheduling policy associated with the historical state is determined. Based on the historical scheduling policy, a computing task to be executed by the computing device is allocated to at least one other computing device in the computing system.