改进CNN模型在物联网数据通信计算中的研究  

Research on improving CNN model in internet of things data communication computing

作　　者：高凤 吴艺妮 褚诗伟 GAO Feng;WU Yini;CHU Shiwei(Information Engineering College,Anhui Lvhai Vocational College of Business,Hefei 231200,China;School of Electronic Information Engineering,Anhui University,Hefei 231200,China)

机构地区：[1]安徽绿海商务职业学院信息工程学院,安徽合肥231200 [2]安徽大学电子信息工程学院,安徽合肥231200

出　　处：《邵阳学院学报（自然科学版）》2024年第3期11-19,共9页Journal of Shaoyang University：Natural Science Edition

基　　金：安徽省教育厅自然科学重点研究项目(2022AH052854)。

摘　　要：为降低传输功耗和解决网络带宽不足的问题,传统方法是通过减小数据体积和优化计算方法来实现。文中提出一种结合边缘计算的改进CNN神经网络模型方法。通过仿真实验测试,证明了其在功耗、队列平均和整体性能方面优于云计算和传统边缘计算。实验表明,优化后的算法能量消耗比云计算低约15 W,比传统边缘计算低约18 W,整体性能得到显著提高。所提改进算法模型能有效提升物联网数据通信管理中的性能,并解决数据采集终端的高功耗问题。To reduce transmission power consumption and solve the problem of insufficient network bandwidth,traditional methods achieve this by reducing data volume and optimizing computational methods.In this paper,an improved CNN neural network model combined with edge computing was proposed,and the simulation experiments verified that it was superior to cloud computing and traditional edge computing in terms of power consumption,queue average and overall performance.Experiments show that the power consumption is about 15 W lower than cloud computing and about 18 W lower than traditional edge computing,and the overall performance was significantly improved.The proposed improved algorithm model enhances the performance of data communication management in the internet of things and solves the high power consumption problem of data acquisition terminals.

关 键 词：边缘计算 CNN模型 物联网数据 通信管理 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]