Research On Task Scheduling And Resource Management In Edge Cloud Computing

With the continuous popularization of cloud computing technology and rapid development of mobile network technology in recent years,mobile internet users who possess mobile smart terminals has exploded.The popularity of new mobile application scenarios such as facial recognition,car networking and enhanced virtual reality put great demands on the bandwidth capacity and network latency.However the wide-area cloud environment has its limitation because of the distance between computing node and user is large,which is likely to cause great delay.Therefore the concept of terminating the forwarding path of the data and processing tasks demanding bandwidth and delay to wireless side has become an increasingly common understanding in the industry.The Mobile Edge Computing(MEC),which meets the demand of front concept becomes the reality and key technology to improve user experience in future network.MEC is a new type of mobile cloud computing architecture that provides computing capability at the edge of any network access to mobile users.MEC meets the requirements of ultra-low latency,ultra-low power,ultra-reliable,ultra-dense services and solves the problems of long waiting time when exchanging data with wide-area public cloud and limitation of time and space in cloudlet.The research content of this paper is how to schedule computing tasks and manage computing resources in MEC network environment to maximize the operational efficiency of the entire MEC working group which contains both edge cloud,mobile terminal,reduce energy consumption and improve user experience.Firstly,the paper analyzes the current situation and existing problems of task scheduling and resource management of mobile users,then summarizes the related problems and research status.The work of front lays the foundation for putting forward and expanding the key points of dissertation.Secondly,this paper establishes an optimized system model,including network model,communication model,calculation model and cache model based on the actual situation of mobile user task scheduling and resource management.After analyzing the advantages and disadvantages of traditional centralized optimization algorithm,the paper proposes several optimization performance indicators and utility functions for task scheduling and resource management respectively.Through the introduction of game theory formulation,mathematical basis of game theory,augmented Lagrange matrix and alternating direction method of Multipliers and other related mathematical tools,the dissertation proposes the distributed task scheduling algorithm based on game theory and the distributed resource management algorithm based on ADMM iteration with the help of existing mathematic algorithms such as potential function analysis,Nash equilibrium conditions,variable relaxation and variable recovery.The algorithm is applied to simulation experiments,the experimental results show the effectiveness of the algorithm to meet the optimization objectives.Finally,this paper analyzes the actual business requirements of task management in an explosion-proof testing center and designs both mobile terminal and server terminal of task management system by using layered architecture respectively.After the proposed distributed algorithm integrated into the system,the paper introduces the main processes and operations of system briefly.From the perspectives of practical application and practical effects,the system which is simple in operation,clear in interface,logical in logic,clear in function can provide auxiliary decision support for task scheduling and resource management for staff member.