Task Scheduling And Resource Management Based On Edge Intelligenc

In the era of cloud computing,enterprises typically transmit data to cluster-based cloud computing centers for processing.In the past,this work model could largely meet the business needs of traditional industries.In recent years,with the development of 5G technology and artificial intelligence,various novel applications have emerged,and the generation of data has accordingly increased significantly.This requires the network to give rapid and precise responses to user task requests,which is difficult for traditional cloud computing models to achieve.Therefore,the marginalization of network computing facilities is a new trend in the future network computer architecture technology.This article combines an analysis of common scenarios in edge networks and studies the resource allocation and task scheduling problems in edge networks.The contributions of this article are as follows:(1)For the resource allocation problem in edge networks,this article proposes an edge network bandwidth adaptive allocation framework based on deep reinforcement learning.We use a video transmission scenario of single-server multi-user in edge network to design an adaptive streaming framework based on network bottleneck bandwidth driven by the goal of providing a fair network bandwidth environment for multiple users in edge network,and maximize the quality of service experience of users as much as possible.Finally,simulation experiments demonstrate that our model has good convergence,and shows good training speed and accuracy in comparison with other literature models.(2)For the task scheduling problem in edge networks,this article proposes a smart agent task scheduling framework based on AC algorithm.We regard an edge server as a smart agent with computing capabilities,which contains an actor network and a critic network.We construct the smart agent model using the AC algorithm and study the task scheduling cooperation schemes between edge computing nodes.Finally,multiple groups of control experiments demonstrate that the smart agent scheduling framework proposed in this article has good performance in complex environments,and thereby ensures the quality of service experience of users.