Edge Server Task Scheduling via Deep Q-Learning Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing task offloading methods in edge network computing fail to adapt to time-varying network conditions, leading to substantial data transfer and processing delays due to congestion and lack of load balancing across cloud servers.
Innovation Solution
The implementation of a deep reinforcement learning based task scheduling system, where edge servers use a deep Q-learning neural network with experience replay to dynamically select cloud servers for task offloading, thereby adapting to changing network conditions and optimizing task distribution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If existing task offloading methods are used in edge network computing, then task distribution can be achieved, but data transfer and processing delays increase due to congestion and lack of load balancing across cloud servers
Solution Approach 1:
The patent implements dynamic task offloading decisions using deep reinforcement learning that continuously adapts to changing network conditions. The system dynamically selects which cloud servers to offload tasks to, based on real-time network state, server load, and congestion levels, rather than using static offloading rules. This dynamic approach resolves the contradiction by making the system adaptable to time-varying conditions while minimizing delays through intelligent, context-aware decision-making.
Solution Approach 2:
The patent employs feedback mechanisms where the deep reinforcement learning model learns from past offloading decisions and their outcomes. The system continuously monitors network conditions, task completion times, and server responses, using this feedback to improve future offloading decisions. This feedback loop enables the system to adapt to changing network conditions and avoid congested servers, thereby reducing data transfer and processing delays while improving overall system adaptability.
2Loss of time
If deep reinforcement learning based task scheduling is implemented, then latency for computation-intensive tasks is reduced, but system complexity increases
Solution Approach 1:
The patent introduces a deep Q-learning neural network as an intermediary component that handles the complex decision-making process. Instead of implementing complex scheduling logic throughout the entire system, the DQN serves as a centralized intelligent mediator that receives system state information and outputs optimized offloading decisions. This intermediary approach reduces overall system complexity by concentrating the intelligent decision-making in a specialized component while keeping the rest of the system relatively simple.
Solution Approach 2:
The patent replaces traditional mechanical or rule-based task scheduling mechanisms with a learning-based neural network system. Instead of using predefined algorithms or manual configuration for task offloading decisions, the system uses a deep Q-learning neural network that automatically learns optimal strategies through experience. This substitution reduces the need for complex manual tuning and system configuration, thereby managing complexity while achieving lower latency through adaptive intelligent decision-making.
Data Source
AI summary
An edge network computing system includes: a plurality of terminal devices; a plurality of edge servers connected to the terminal device through an access network; and a plurality of cloud servers connected to the plurality of edge servers through a core network. Each edge server is configured to: receive a plurality of computing tasks originated from one of the plurality of terminal devices; use a deep Q-learning neural network (DQN) with experience replay to select one of the plurality of could servers to offload a portion of the plurality of computing tasks; and send the portion of the plurality of computing tasks to the selected cloud server and forward results of the portion of the plurality of computing tasks received from the selected cloud server to the originating terminal device.