基于随机森林的车载CAN总线异常检测方法

Anomaly Detection Method for In-Vehicle CAN Bus Based on Random Forest

针对目前车载网络的信息安全问题,在控制器局域网(CAN)总线异常检测方法的基础上,提出一种基于随机森林模型的CAN总线报文异常检测方法.首先用采集的大量正常和异常报文数据构造随机森林模型,并进行一系列的参数调整;然后将待检测的CAN总线报文输入到对应ID的随机森林模型中;最后通过模型完成报文正常或异常的分类.仿真实验结果表明,该模型能有效检测出总线上的异常数据,提升了汽车运行的安全性.

Aiming at the information security problems of in-vehicle network,on the basis of anomaly detection method of the controller area network（CAN）bus,we proposed an anomaly detection method for CAN bus message based on the random forest model.Firstly,a large number of normal and abnormal message data were used to construct a random forest model and perform a series of parameter adjustments.Secondly,the CAN bus message to be detected was input into a random forest model of the corresponding ID.Finally,a classification of the normal or abnormal message was completed by the model.The results of simulation experiment show that the model can effectively detect the abnormal data on the bus,and improve the safety of the vehicle operation.