摘要
随着互联网、云计算和大数据的发展以及5G通信的普及,出现了很多由联网计算机控制的智能网联汽车.这种发展为人们的生活带来很大便利,但与此同时,这种与车紧密相关的网络的安全性也存在着许多问题,近年来已经证明了它们存在的许多攻击漏洞.因此,针对这种安全隐患,本文所提出的模型,通过使用多种机器学习算法来对入侵车载CAN网络进行入侵检测.首先会介绍几种常见的针对CAN网络的攻击方式,然后基于报文数据特征,分别使用Adaboost,KNN,SVM三种算法实现分类.最后,使用基于三种算法的检测模型对真实采集到的数据分别进行测试,对比检测结果,得到了三者中能实现更高准确率的算法Adaboost,准确率达到了96.22%.
With the development of the internet,cloud computing and big data,and the popularity of 5G communications,there have been many intelligent networked cars controlled by networked computers. This kind of development has brought great convenience to people′ s lives,but at the same time,there are many problems in the security of such a car-related network. In recent years,many attack vulnerabilities are proved. Therefore,for this security risk,the model proposed in this paper uses a variety of machine learning algorithms to intrusively detect intrusion on-board CAN networks. Firstly,several common attack methods for CAN networks will be introduced. Then,based on the characteristics of the message data,three algorithms,Adaboost,KNN and SVM,are used to classify. Finally,using the detection model based on three algorithms to test the real data separately,and comparing the detection results,Adaboost achieves higher accuracy with an accuracy rate of 96.22%.
作者
谢浒
莫秀良
王春东
XIE Hu;MO Xiu-liang;WANG Chun-dong(School of Computer Science and Engineering,Tianjin Key Laboratory of Intelligence Computing and Novel Software Technology,Tianjin University of Technology,Tianjin 300384,China)
出处
《天津理工大学学报》
2020年第2期32-37,共6页
Journal of Tianjin University of Technology
基金
天津市特派员项目(18JCTPJC51000).
关键词
CAN网络
网络攻击
入侵检测
机器学习
CAN network
network attacks
intrusion detection
machine learning
