Research On Edge-environment-based Multi-workflow Scheduling With Unstable Infrastructural Performance

In recent years,with the rapid development of advanced communication technology represented by 5G,the edge computing paradigm has gradually become an important cornerstone for boosting emerging mobile applications,innovating human-computer interaction,accelerating big data governance of the Internet of Things,and improving industrial ecology from concept to reality.Edge computing is a complex system that creates a new framework to support tight integration between wireless networks and cloud computing.Due to the characteristics of edge computing communication and computing collaboration,the research and development of edge computing is faced with a wide range of challenges,and it is also a research hotspot in academia and industry at home and abroad.Edge computing is an open,elastic,collaborative system where data,computation,storage,and applications are close to data sources.It can offload computing tasks from mobile terminals to the edge of the network for processing,and can provide more efficient storage,computing,network services,etc.A key problem of workflow scheduling in edge computing environments is how to reduce the cost and response time while the performance and quality of service requirements meet the service level agreement(SLA).Ensuring that users perceive the application's quality of service remains a challenge as applications continue to suffer from negative effects,such as network congestion,long message latency,and reduced edge server coverage due to battery depletion.In this thesis,the problem of multi-workflow scheduling is studied,and a method of multi-workflow scheduling with proximity constraint in edge computing environment is proposed,aiming at economic benefit.The proposed method is designed to minimize the cost of edge computing while satisfying the completion time of user-specified workflow.Discrete firefly algorithm is used to generate scheduling policies.A large number of case studies are conducted based on several well-known scientific workflow templates and real datasets of edge server locations.The experimental results clearly show that the proposed method is superior to the traditional methods in terms of cost and completion time.Meanwhile,when the server is close to the end user in the edge computing environment to provide low latency and low energy services,the computing and radio resources used are limited,so this requires quality of service assurance and efficient task scheduling methods and strategies.In order to solve the problem of multi-workflow scheduling in edge environment,this thesis also proposed a multi-workflow scheduling method based on probabilistic Qo S awareness.The proposed approach uses the Qo S aggregation model based on the probabilistic mass function and the discrete firefly algorithm to generate the multi-workflow scheduling plan.In order to prove the effectiveness of the proposed method,this thesis studies the different types of workflows,edge servers' performance and real datasets of edge server locations.Experimental results show that the proposed method is superior to other methods in terms of completion time,cost and global Qo S probability.