移动通信2024，Vol.48Issue(12)：129-133,5.DOI:10.3969/j.issn.1006-1010.20240523-0002
基于联邦学习的物联网感知数据安全方法研究
Research on IoT Perceived Data Security Method Based on Federated Learning
包超明 1包森成 1刘亦哲1
作者信息
- 1. 中国移动通信集团浙江有限公司,浙江 杭州 310000
- 折叠
摘要
Abstract
To address the issue that the accuracy of current IoT perception data security methods is easily affected by non-independent and identically distributed phenomena,a model suitable for IoT perception data security is constructed.This model applies Bayesian algorithm to the federated learning framework,learns the probability distribution of local model parameters,quantifies the uncertainty of local model parameters using Bayesian algorithm,which solves the negative impact of unreliable local model parameters in the global parameter aggregation process.Simulation results show that the research method of IoT perception data security based on federated learning has faster convergence speed and higher accuracy than traditional FedAvg,FedProx,and SCAFFOLD algorithms.关键词
联邦学习/物联网感知/数据安全/贝叶斯Key words
federated learning/IoT awareness/data security/Bayesian分类
信息技术与安全科学引用本文复制引用
包超明,包森成,刘亦哲..基于联邦学习的物联网感知数据安全方法研究[J].移动通信,2024,48(12):129-133,5.
