Research On Intrusion Detection System Of Vehicle CAN Network Based On Neural Network

Vehicle networking is to connect cars to the Internet which is to break the information barrier between cars and the Internet.It has brought great variables to people's daily life and traffic management.However,due to the development of vehicle network architecture is limited to the early time of the Internet and the auto industry,which makes the automobile network exists a huge security problem.In order to deal with these threats,intrusion detection system for a vehicle CAN(Controller Area Network)network is proposed which is based on neural network.This paper have done following research.(1)Understand the research of the current domestic researchers on the safety of vehicle network.According to the research nature,it is divided into four categories: encryption and authentication,intrusion detection and firewall and security framework.By analyzing the feasibility of these security means,it is concluded that the encryption and authentication methods,firewall technology and security framework are limited by the current in-vehicle network environment has no practical help for CAN bus security.Therefore,research on intrusion detection system which is base on neural network is very import.(2)The characteristics of CAN network are including electrical characteristics,communication principles,protocol types and security risks.Summing up the external and internal attacks that may occur in the car.The intrusion detection system for invehicle CAN network based on neural network is designed which have ability to defend against injection attacks.Denial of service attacks,replay attacks,and tamper attacks.(3)Analysis of the frequency of data transmitted by each ECU(Electronic Control Unit)on the CAN network.Using PCA-BP(Principal Component AnalysisBack Propagation)neural network to fit the relationship between the frequency of ECU sending data and the current vehicle state.PCA is use for principal component extraction,reduce the dimension of input data of neural network and enhance the convergence rate(4)Analyze the hidden relationship between ECU data.For example,there is a relationship between the speed of the rotational speed engine and the opening of the collector valve.Using GA-BP(Genetic Algorithm-Radial Basis Function)neural network to fit the correlation of ECU data.Finally,using this correlation to judge the security of each data in the vehicle network...