Edge Task Scheduling Using Overhead-Aware Computing Power Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing edge collaboration implementations have not adequately considered scheduling overhead, leading to additional network costs and reduced user experience quality in collaborative edge computing, particularly in multi-region inter-connected edge networks with unpredictable task requests and varying system states.
Innovation Solution
A task scheduling method that determines a scheduling strategy for each service in a multi-region inter-connected edge network based on scheduling overhead, prediction information, and network state, using a computing power network graph to guide task requests across network access points, optimizing scheduling within each cycle to balance short-term performance and long-term constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If task scheduling is performed among edge nodes in collaborative edge computing, then the handling capacity for service requests is improved, but the network cost and scheduling overhead increase
Solution Approach 1:
The patent transforms the discrete scheduling decision into a continuous parameter optimization problem by using scheduling probabilities (continuous values between 0 and 1) instead of binary decisions. This allows for fine-grained control of task forwarding behavior, enabling the system to adjust scheduling intensity and targets dynamically to minimize network costs while maintaining service handling capacity.
Solution Approach 2:
The patent implements feedback mechanisms by using historical scheduling overhead information and current network state to dynamically adjust scheduling strategies. The controller continuously monitors scheduling outcomes and uses this feedback to optimize future scheduling decisions, thereby reducing network costs while maintaining productivity.
2Loss of energy
If scheduling strategy is optimized for long-term cost control, then the average cost is reduced, but the response time for individual tasks may increase
Solution Approach 1:
The patent makes the scheduling strategy dynamic by adjusting scheduling probabilities based on real-time network state and historical performance. This dynamic adjustment allows the system to respond to changing conditions, balancing cost optimization with response time requirements adaptively rather than using a fixed strategy.
Solution Approach 2:
The patent performs preliminary actions by pre-calculating scheduling strategies based on historical data and prediction information before tasks arrive. This allows the system to prepare optimal scheduling decisions in advance, reducing both cost and response time when actual tasks need to be scheduled.
3Adaptability or versatility
If scheduling decisions are made based on current network state, then the adaptability to varying conditions is improved, but the complexity of scheduling control increases
Solution Approach 1:
The patent segments the complex scheduling control into two parts: a central controller that performs heavy computation using prediction information and historical data, and distributed edge nodes that execute simplified scheduling decisions based on received strategies. This segmentation reduces the complexity burden on individual components while maintaining overall adaptability.
Solution Approach 2:
The patent introduces a central controller as an intermediary that handles the complex analysis and strategy generation, then communicates simplified scheduling strategies to edge nodes. This intermediary absorbs the computational complexity while enabling adaptive scheduling at the distributed edge nodes.
Data Source
AI summary
The present disclosure provides a task scheduling method, electronic device, and storage medium. The method is performed by a central controller in a multi-region inter-connected edge network, including: receiving, from each of a plurality of network access points, first scheduling overhead, prediction information for task arrival, resource allocation information, and network status information; generating a computing power network graph based on the prediction information for task arrival, the resource allocation information, and the network state information of the plurality of network access points; determining a scheduling strategy for a next scheduling cycle based on the first scheduling overhead of the plurality of network access points and the computing power network graph; and sending the scheduling strategy to each of the plurality of network access points before the next scheduling cycle begins.