Research Of FlexRay Information Anomaly Detection Based On Message Perception Method

With the development of automobile intelligence,the number of electronic control units has increased,and the functions of automobiles have become more and more extensive,more like a mobile computer.In order to realize many functions such as automobile interconnection,human-vehicle interconnection,user analysis,etc.,the in-vehicle network is no longer a closed and safe space.Automobiles need to interact with the outside frequently.While the interface with external networks has increased significantly,it also increases the possibility of intrusion.The intruded car can be controlled by hackers,such as wipers,steering wheel,speed,etc.,which can affect driving at a light rate,and cause a traffic accident in a serious way.With the increase in the number of interfaces between automobiles and the outside world and the increase in electronic components,the research on automobile safety has gradually shifted from mechanical safety to bus information safety.Among many vehicle bus networks,FlexRay bus network,as an emerging vehicle bus network with high transmission rate,high fault tolerance,and high real-time performance,has been used in chassis control and other safety-related systems directly by automobile brands such as BMW and Audi.The safety of FlexRay vehicle bus network is of paramount importance.The vehicle-mounted network is different from the traditional computer network,so it is necessary to conduct research according to the characteristics of the vehicle-mounted network to effectively improve data security.This paper proposes a design for detecting anomalies in vehicle FlexRay bus network information based on message perception.This article introduces the information characteristics of the vehicle-mounted FlexRay bus network,and analyzes the existing vehicle-mounted bus network security strategy.According to the characteristics of the vehicle bus network information,the message perception design is carried out.The FlexRay information generation design is proposed to solve the problem that FlexRay bus network data cannot be directly read out through the OBD interface,and to realize the data problem of FlexRay related experiments.Combining the message perception results and SAE standards,the FlexRay information is generated by simulation.The method of randomly selecting locations and breaking the correlation of information is used to generate abnormal information that meets the actual situation,replace the correct data,and generate an information data set that can be used for anomaly detection.Build a convolutional neural network to detect the relevance of information,and adjust the parameters to achieve a high detection rate.Perform fault-tolerant control on the detected abnormal information,and propose a fault-tolerant method based on the first X pieces of information.Finally,use software such as Spyder,CANoe,Codewarrior,and the hardware development board to build an experimental framework model,complete joint experiments,and verify the operability of the design.Experiments show that this design has a detection accuracy rate of 99% for abnormal information in the vehicle-mounted FlexRay bus network.