Research On New Architecture Of Edge Computing Based On Stratified Sampling And Method Of Computing Task Replications Distributiton

At present,the development of Internet of Things technology is very rapid.The Internet of Things not only penetrates into people's daily life as an infrastructure,but also plays an important role in industrial manufacturing.With the growing scale of the Internet of Things,the centralized computing model has been unable to adapt to large-scale Internet of Things data processing and analysis.Therefore,how to efficiently process the real-time data stream generated by the large-scale Internet of Things has become an urgent problem to be solved.In response to this problem,this paper proposes a real-time data stream processing framework for the Internet of Things based on Edge Computing-Approx ECIo T.The core idea of the framework is Approximate Computating.Through stratified sampling of the data stream,the size of the sample in the layer can be adjusted adaptively according to the error size,so as to realize efficient data processing and analysis.In addition,a method of task's replications distribution oriented to edge computing network load balancing is also proposed.By realizing the optimal allocation of replication resources of edge nodes and load balancing-based replications distribution,the computing efficiency of user tasks is improved and the calculation delay is reduced.The main work and innovations of this paper are as follows:Firstly,for the problems of the traditional Internet of Things being unable to handle real-time data streams,excessive load pressure on cloud service centers,and low computing efficiency,this paper designs and implements the real-time data stream processing and analysis architecture.The Approx ECIo T architecture takes into account the resource finiteness of edge nodes in the Internet of Things architecture.When processing and analyzing these large amounts of data streams,the limited resources should be used as efficiently as possible.In this architecture,in order to ensure that the sampling calculation results meet the user's requirements,an error adjustment mechanism is introduced.When a node is performing a calculation task,if the calculation result does not meet the accuracy requirements given by the user,then the capacity of each layer of the sample is adjusted,and then the sampling calculation is repeated.Finally,in order to improve the resource utilization efficiency of nodes in the Internet of Things system,Approx ECIo T also introduces self-adjusting computations to realize the reuse of the results of some computation tasks.The experimental results show that Approx ECIo T can still obtain high-accuracy calculation results even with limited edge node resources.Secondly,due to the current Internet of Things computing architecture,user task computation efficiency is low,and the response time is long,this paper designs and implements a task processing algorithm based on edge computing network load balancing,which makes full use of the resources of network edge nodes to improve computing efficiency and reducing the load pressure of the cloud service center.Firstly,the algorithm divides the user tasks of the cloud service center into multiple subtasks,and distributes these subtasks to the edge nodes in the network through the roulette algorithm,so that nodes with more neighbor nodes take on more subtasks.Each edge node replicates multiple copies of the subtasks undertaken by each edge node.In this process,the optimal allocation of replications resources is realized.At the same time,in order to ensure that task replications can get a timely response,a new load balancing strategy is proposed.Finally,through the use of simulated data stream and real data stream for experimental testing,the experimental results show that the Io T task processing strategy proposed in this paper has a significant improvement in efficiency compared with other task computing mode.