摘要
针对车载控制器局域网络(controller area network,CAN)总线入侵检测准确率低与时效性差的问题,通过分析总线中入侵数据帧的特点,提出了基于门控循环单元(gated recurrent unit,GRU)的入侵检测方法。该方法搭建了由5层神经网络构成的入侵检测模型,以真实汽车采集的CAN数据为基础构造洪泛攻击、重放攻击、模糊攻击和虚拟节点攻击数据,提取出具有11个特征的特征向量序列用于模型的训练和测试。实验验证了模型参数对检测结果的影响,研究了二分类检测和多分类检测的准确率与时间开销。结果表明:该方法在二分类和多分类检测中的精度为99.9816%和99.8942%,召回率分别是0.9999和0.9991,达到与长短期记忆(long short-term memory,LSTM)模型相当的检测精度,并且具有更短的训练和检测时间。本文方法提高了入侵检测的时效性和可靠性,对保障汽车安全意义重大。
To solve the problem of low accuracy and timeliness for in-vehicle CAN bus intrusion detection,an intrusion detection method based on gated recurrent unit(GRU)was proposed by analyzing the features of intrusive data frame.The proposed method built an intrusion detection model consisted of five-layer neural network,and constructed DoS(denial of service)attack,replay attack,fuzzy attack and virtual node attack data based on real CAN data,used for extracting 11-feature vector sequences.The impact of model parameters on detection results was analyzed,the accuracy and efficiency of binary and multiclass classification detection was tested as well.The results show that the accuracy of the proposed method reaches 99.9816%and 99.8942%in binary and multiclass classification,the corresponding recall rates are 0.9999 and 0.9991,respectively.The proposed method is equal to LSTM(long short-term memory model)in accuracy,and has shorter training and detection time.The proposed method improves the efficiency and robustness of intrusion detection,which has great significance for improving the safety of vehicles.
作者
许秀锋
蒲家坤
周爱国
于江洋
李振雨
XU Xiu-feng;PU Jia-kun;ZHOU Ai-guo;YU Jiang-yang;LI Zhen-yu(School of Mechanical Engineering, Tongji University, Shanghai 201804, China)
出处
《科学技术与工程》
北大核心
2021年第16期6786-6793,共8页
Science Technology and Engineering
基金
国家重点研发计划(2016YFB0100902)。
