Research On Performance Feature Detecti On And Task Scheduling Of Edge Nodes Based On Heterogeneous Environment In Edge Computing

With the expansion of the scale of the.Internet and the development of tec hnology,more and more equipment terminals are connected to the network,an d data is exploding.The traditional cloud computing centralized network archit ecture has become the bottleneck of massive data processing,and the edge co mputing architecture that migrates data to the edge-side network for processing has emerged.The edge node carries the computing and storage capabilities on the edge side which composed of different heterogeneous physical devices and hardware resources.However,in the process of offloading edge computing tas ks,the resource characteristics and performance of heterogeneous edge nodes ar e difficult to be accurately perceived,and there is a lack of performance detect ion methods for heterogeneous edge nodes in real work environments.For edg e computing,task scheduling and execution are the key factors affecting perfor mance,while the optimization of task scheduling is often linked to the node-ta sk feature relationship.Therefore,in edge computing,the lack of edge node re source performance characteristics has become an important factor restricting th e development of edge node task scheduling algorithms and improving edge co mputing performance.To solve the above problems,this paper first proposes a resource-aware pe rformance feature detection method for heterogeneous edge nodes in the edge n etwork.This method proposes a single intensive load model to decouple the co mplex impact between resources,establish a resource utilization matrix,extract the relative performance characteristics of the same type of resources at the ed ge nodes based on the resource relative performance quantization algorithm,an d use relevant verification methods to ensure the characteristic conclusions corr ectness.At the same time,this method combines the resource utilization matrix and the task load model to give a correlation analysis of the internal resource efficiency of the edge nodes,and finally realizes the multi-dimensional and sy stematic feature detection and description of heterogeneous edge nodes.At the same time,based on the above feature detection results,this paper designs an edge node task scheduling algorithm,defines task scheduling perfor mance indicators,combines task scheduling weights based on node resource per formance characteristics,designs a task scheduling mechanism based on the ed ge network structure,and provides a task scheduling algorithm overall design with building the corresponding model.Finally,we evaluate and analyze the performance of the edge node task sc heduling algorithm based on feature weights through experiments.The performa nce differences of the optimal scheduling algorithm in the two dimensions of t ask execution time and resource use efficiency compared to the control group are 24.6%and 248%,respectively.