Low-delay task scheduling method for a dynamic fog computing network

Abstract

The invention provides a low-delay task scheduling method for a dynamic fog computing network, which is characterized in that a help node in the network cannot broadcast own node state information inreal time, such as task queue information and sharable computing resource information, or the help node cannot respond to requests of the information in real time. And when a task unloading requirement exists each time, the task node needs to make an unloading decision in real time, and an unloading task is selected from the current candidate help nodes. Due to the fact that the task nodes have unknown states for the help nodes and the tasks have time delay requirements, the task nodes need to well unload experiences from previous tasks for learning, and judgment is provided for current decision making. The invention provides a one-to-many task unloading algorithm based on an online learning method for non-dynamic and dynamic changing fog computing or edge computing networks, and the method can greatly reduce the energy expenditure caused by information transmission in the network and prolong the service time of task nodes and help nodes.

Description

technical field [0001] The invention belongs to the field of computing communication networks, and in particular relates to a task offloading algorithm aiming at reducing the average task offloading delay. Background technique [0002] With the rise and development of technologies such as intelligent Internet of Things, 5G, and artificial intelligence, processing massive and diverse data and meeting ultra-low service latency requirements have become more and more urgent problems to be solved. The traditional centralized cloud computing architecture has a large delay due to the long distance between the terminal device and the cloud server, and it is difficult to independently meet the needs of delay-sensitive services; access. In this context, fog computing came into being. With its distributed framework and low-latency services, fog computing is expected to become a key technology supporting future intelligent IoT, 5G and artificial intelligence applications, and has rece...

AI Technical Summary

Problems solved by technology

Continuously broadcasting and monitoring node status information in real time will generate a large amount of energy consumption and affect the service life of the equipment. In the future ultra-large-scale system, this problem will become more prominent

[0006] In the actual situation, nodes will dynamically enter and exit the network, and the shareable computing resources will also change with the situation. At the same time, it is hoped to minimize the energy consumption in the network and prolong the service life of the equipment. Such a problem that is most close to the actual demand and needs to be solved , but few scholars have studied

Images