基于单分类支持向量机的CAN总线异常检测方法

Abnormity Detection Method for In-Vehicle CAN Bus Based on One-Class SVM

为提高智能客车的网络安全性,提出一种基于单分类支持向量机模型的CAN总线报文异常检测方法,根据智能客车CAN总线的报文数据域特性,分析攻击对数据域产生的影响,将CAN报文的数据域提取成8个训练特征,以大量的行驶数据作为训练集和测试集,通过随机和仿真方式生成异常数据,采用交叉验证的方式对参数进行调整。试验结果表明,该模型能有效检测出异常数据,提升了智能客车的行驶安全性。

In order to improve the cyber security of intelligent bus, this paper proposes a CAN bus messageabnormity detection method based on One-Class Support Vector Machines(OCSVM) model. According to the characteristicsof message data domain of intelligent bus CAN, this paper analyzes the impact of attack on data domain, and extracts thedata domain of CAN message into 8 training features. A large number of driving data are used as training set and test set,which generate abnormal data by random and simulation mode, and the parameters are adjusted by cross validation. Thesimulation results show that the model can effectively detect the abnormal data and improve the driving safety of intelligentbus.