Research On Edge Cloud Collaborative Design And Task Scheduling Algorithm Of Internet Of Things

Delay-sensitive Io T scenarios require Io T platforms to provide low-latency,high-quality intelligent services.The computing centery of the Io T platform with a centralized architecture is far away from the terminal device,and the data transmission delay is high.The use of edge computing technology to deploy services on the near device side can solve the problem of high latency in the cloud center,but the decentralized deployment of edge clouds makes it difficult to coordinate service management At the same time,a single edge cloud computing resources are limited,and it is difficult to independently handle large-scale computing tasks.In response to the above problems,this paper proposes an Io T edge cloud collaborative architecture suitable for delay-sensitive application scenarios.Based on the proposed architecture,it studies online scheduling algorithms for large-scale delay-sensitive tasks.Aiming at the problem of decentralized deployment of edge cloud and distributed service management difficulties,a microservice-based edge cloud service management method is proposed to realize unified supervision and dynamic configuration of edge cloud services;addressing the problem of difficult service cross-edge cloud collaboration,based on g RPC technology to design diversified call methods and standardized service call interfaces to achieve efficient collaboration between services;based on RBAC model and OAuth2.0 technology,design service access control mechanisms to achieve security control of basic service components of the Internet of Things Sharing: Aiming at the problem that the edge cloud is difficult to handle large-scale computing tasks independently,an edge cloud collaborative task scheduling mechanism is proposed.By flexibly configuring offline scheduling strategies for the task scheduler in the edge cloud,tasks are scheduled to the cloud computing center or other loads The lighter edge cloud realizes efficient collaborative processing of tasks.Aiming at the scenario where large-scale delay-sensitive tasks are randomly generated,an online task scheduling algorithm based on completion time estimation is proposed,which comprehensively considers factors such as transmission delay,waiting delay,calculation time,and resource contention to minimize the average task completion time For the goal,scheduling decisions are made based on the estimated task completion time of candidate computing nodes.Experiments show that the average task completion time of the proposed scheduling algorithm is reduced by28.3% and 12.6% compared with the minimum delay priority algorithm and the minimum waiting queue priority algorithm;For online task scheduling scenarios with time constraints,a two-stage multi-path scheduling algorithm is proposed.In the first stage,tasks are pre-allocated to candidate computing nodes,and the improved FIFO algorithm is used to check the feasibility of scheduling and calculate the slack time;In the second stage,in order to avoid the problem of load imbalance caused by information delay and resource contention,the corresponding scheduling probability is set for the feasible scheduling node according to the proportion of the slack time,and the scheduling decision is made according to the probability.Experiments show that the missed task deadline of the proposed two-stage scheduling algorithm is 15.4% and26.7% lower than the EDF(Earliest Deadline First)algorithm and the task scheduling algorithm based on completion time estimation,respectively.Based on the proposed architecture and method,the edge cloud collaborative Io T platform was designed and implemented,and it was applied to projects such as "smart building","smart sentry",and "smart agriculture" to verify the usability of the research method.