Task Scheduling Stratrgy In End-edge-cloud Collaborative Network

With the advent of the era of big data,the amount of data computation generated by mobile devices has increased exponentially,resulting in a large number of computation-intensive and time-delay sensitive services.Due to limited resources and computing power of mobile terminal devices,a large number of tasks cannot be timely processed,which affects user experience.At the same time,the traditional business can no longer meet the needs of users,and the user business presents a diversified trend.Users’requirements for services are becoming more customized and intelligent.In order to deal with the challenges brought by massive data processing and diversified business demands,the edge-cloud collaborative network has come into people’s view.As the key technologies of edge-cloud collaborative network,task scheduling and resource allocation technology schedules tasks generated by mobile devices to compute nodes in the network for processing and allocation of network resources.It can stimulate the synergy of edge-to-edge cloud and improve the utilization rate of network resources,so as to better meet the diversified needs of users’tasks.How to design reasonable and efficient task scheduling and resource allocation strategies in the edge-cloud collaborative network environment to meet users’ task requirements is a hot issue in current research.Therefore,this paper studies the task scheduling and resource allocation strategy under the end-edge-cloud collaborative network,and its main contributions are as follows:1.To solve the problem that mobile terminal devices cannot meet users’ demand for high-speed service processing,a heuristic task scheduling and resource allocation strategy oriented to low delay is proposed and implemented in this paper.In this paper,the task processing delay and the computing resource allocation form of Multi-access Edge Computing(MEC)server in the classical edge-cloud collaborative network were modeled and analyzed with the minimum total system delay cost as the optimization objective.The optimization problem was divided into two sub-problems,the optimal resource allocation problem and the task scheduling problem under the fixed task scheduling strategy.KKT(KarushKuhn-Tucker)condition and heuristic based task scheduling algorithm were used to solve the problem respectively,and the total delay cost of the system was minimized.The simulation results show that compared with the common random scheduling and greedy scheduling algorithms,the proposed algorithm reduces the total system delay cost by about 43.5%and 35%,respectively,taking the number of tasks as an example.At the same time,when the number of tasks,MEC server resources,the amount of computing task input data and network bandwidth resources change dynamically,the proposed scheduling strategy can still maintain the optimal performance.2.In view of the diversified user needs,different types of businesses have different quality of service requirements,different types of businesses have different computing workload,business data size and delay requirements,this paper proposes and implements a task scheduling strategy oriented to business priority.First,this paper uses analytic hierarchy process to prioritize different types of services according to their computational workload,business data size and latency requirements.Then,a task scheduling and resource allocation algorithm based on branch and bound is proposed to schedule different kinds of tasks generated by terminal devices and allocate computing resources and bandwidth resources of MEC server.Due to the exponential complexity and low practicability of this algorithm,this paper continues to propose a lowcomplexity linearization-based task scheduling and resource allocation algorithm.The algorithm maximizes the task scheduling efficiency while ensuring the priority of high-priority tasks.The simulation results show that,taking the change of computing resources of MEC server as an example,compared with the exhaustive greedy algorithm,greedy algorithm and random algorithm,the proposed algorithm in this chapter improves the task scheduling utility index by 4.7%,14.3%and 15.5%respectively.3.Based on the proposed low delay oriented task scheduling and resource allocation strategy,a business-oriented visualization platform is designed and implemented.In this paper,the architecture of the visualization platform is designed and the data flow of the whole system is analyzed.The platform relies on a variety of intelligent services such as face recognition,traffic monitoring,image enhancement rendering,distributed model training,etc.,and displays business processing results,network resource data and algorithm performance indicators in the form of a visual interface.By verifying the performance of the strategy,it can be seen that the designed task scheduling strategy can achieve shorter execution delay in the test of different frame extraction speeds and the number of access video streams.