Research on Heterogeneous Optimization Methods for Edge Al Clusters in Smart Campus
With the rise of edge computing and artificial intelligence(AI)technology,the iot so-lutions based on cloud computing cluster have gradually revealed shortcomings.The iot solutions based on edge computing cluster are more and more favored by researchers because of their small size,small inference delay,and flexible cluster deployment.However,most of the current re-search focuses on the computing efficiency of a single AI task in a single edge terminal,which is low.Therefore,this paper applies the existing software and hardware resources of colleges and universities to build a group of edge device clusters that deploy multiple types of reasoning tasks,and studies the edge heterogeneous solutions.This paper proposes a heterogeneous optimi-zation method for edge AI clusters in smart campus,integer linear programming and its solution method,plane division and bound method,to solve the current situation of resource waste caused by unreasonable task scheduling of edge heterogeneous clusters,and lay a solid foundation for the construction of smart campus.
Integer Linear Programming MethodPlane Partition and Bound MethodEdge ComputingEdge Heterogeneous Cluster
NETL
NSTL
万方数据
