Research And Verification Of Key Technologies On Edge Control System For Io T Applications

With the rapid development of cloud computing,artificial intelligence,big data,blockchain,5G and other technologies,various Io T applications are emerging in endlessly.The diverse applications of the Internet of Things put forward higher differentiated demands on the network,such as larger capacity,faster response,more efficient,more flexible,more secure,and so on.However,the cloud computing architecture is relatively fixed,the network system is not flexible enough,and the high latency and network congestion caused by centralized data processing cannot be solved,so it is difficult to perfectly apply to the future Io T services.Therefore,this thesis applies the edge computing architecture to the Internet of Things environment and builds an edge control system.Some key techniques in edge control system are thoroughly studied and validated.This thesis designs an edge control system and verifies related functions.The main work includes:(1)In view of how to reduce the processing delay of small tasks in Io T scenarios,a centralized multi-task scheduling for single edge nodes method.This method firstly designs a multi-level scheduling framework that combines process scheduling and thread scheduling.Process scheduling is targeted at different types of services and uses a preemptive static priority scheduling algorithm.Thread scheduling is aimed at the same type of business with high concurrency.This thesis proposes a task execution urgency factor and designs a thread scheduling algorithm based on dynamic priority.The setting of the urgency factor is determined by the remaining calculation amount during the task deadline.Simulation results show that the proposed multi-level scheduling method plays a significant role in ensuring task timeliness processing and can reduce small task processing delay to a certain extent.(2)In view of how to reduce the processing delay of large tasks in Io T scenarios,this thesis proposes a distributed collaborative computing method with multiple edge nodes based on the idea of "multi-machine collaboration".This method firstly uses asynchronous message queue(MQ)to build a framework for distributed collaborative computing.Through task splitting,multiple nodes can cooperate to serve a task.In the framework's task offloading module,a task offloading decision algorithm based on a multiple 0-1 knapsack model with limited value is designed to minimize the task processing delay with the optimal offloading scheme.The system test results show that the proposed distributed computing framework and offloading decision algorithm can significantly reduce large task processing time.(3)Design and build the edge control system.Firstly,the software and hardware system and edge node management platform are implemented.Then analyzed the communication scenarios of the system's edge layer and perception layer,edge layer and cloud layer,and designed and implemented the service type-driven link selection algorithm and northbound multi-link switching algorithm to ensure the reliability of communication between different levels of the system.Based on the implementation of the above functions,the edge control system can meet the intelligence,scalability,and security requirements of the Internet of Things applications.