Boundary Region Detection Of Internet Of Things Events And Scheduling Optimization Of Complex Tasks Under Edge Computing Environment

The rapid development of Internet of Things(IoT)technology brings opportunities for remote sensing and geospatial information science.All kinds of intelligent devices continuously perceive their surrounding environment,forming a perception network with intensive deployment,mobile coverage,extensive presence and interconnectivity.With the popularity of edge computing,it provides sensing data storage and other application services for IoT terminal devices near the edge of the network,which can further tap the application potential of sensing data of IoT terminal devices.However,due to the limitation of its own hardware capability,it is necessary to balance the energy consumption and working intensity of the equipment in the sensing application based on terminal equipment.On the other hand,when many terminals are connected to the cloud or edge end,it will cause problems such as slow network response.The working mode of IoT devices and the scheduling and processing of tasks by edge network directly affect the efficiency of the system.Aim at event detection on the IoT environment and task scheduling and edge collaboration problems in the edge network environment,this paper studies respectively from IoT continue event boundary detection and complex task scheduling in multi-edge server's collaboration environment.The purposes are optimizing the detection precision of event boundary region,delay and long-term system energy consumption in edge networks.The main research contents include:(1)The detection mechanism of target boundary region based on sensor hybrid deployment mode is proposed.The main steps are as follows: the static sensor is used to perceive the target area in a large range,the boundary area of continuous targets is estimated by spatial interpolation algorithm,and the precise target boundary area is obtained by the exploration of mobile nodes.Experimental results show that the method proposed in this study can effectively improve the detection accuracy of the boundary region,and the heuristic path planning algorithm for mobile nodes can balance the time constraints and energy consumption requirements of the task.(2)A complex task optimization mechanism based on edge node collaboration is proposed.In this method,the complex structure tasks are first split in the cloud to obtain a set of sub-tasks with independent data functions.The directed edge weights between the edge nodes participating in the collaboration are calculated and the execution positions and order of the sub-tasks are generated.The experimental results show that,compared with the traditional cloud computing center model and the general edge computing model,the task cooperative scheduling algorithm based on the minimum spanning tree can greatly reduce the delay of the system processing user requests and improve the user satisfaction.(3)An online scheduling optimization mechanism for time-sensitive complex tasks based on edge server collaboration is proposed.This method is based on the markov decision process to model the problem and schedule the subtasks that belong to different user requests in stages.The approximate method is used to parameterize the value function of the decision action to avoid the dimension disaster caused by the large scale of the system state.The experimental results show that the proposed reinforcement learning-based task scheduling mechanism can optimize the long-term average energy consumption and delay of the system and realize the load balance of the whole network.