Information Security Of VANET And The Research On Anomaly Detection Technology

With the development of the network,the electronic products on the vehicles are constantly enriched,and there are swift growth in remote fault diagnosis,real-time navigation,information,entertainment and other applications.Continuously enriched product function makes the vehicle and Internet companies continue to integrate,vehicles have changed from closed to opened.The concept of intelligent transportation makes the vehicle walking into the intelligent network.Various intelligent devices are continuously integrated and added,and it has become irresistible trend about vehicle and vehicle interconnection,vehicle connected with roadside equipment and vehicle and Internet Interconnection.Vehicular Ad Hoc Network(VANET)has gradually developed.While looking forward to the convenience and benefits brought by intelligent transportation,the emergence of vehicles network security survey,the automotive remote crack experiment showed in international hacker conference remind us,it is urgent to solve the problem of vehicle network security,The development of intelligent transportation should be based on the information security of the whole vehicle linking network.At present,more and more intelligent devices are connected with vehicles.When the entire network security system is not perfect,there must be many security holes in these interconnections.The vehicle information security involves not only personal privacy,personal safety and property security issues,also involves the maintenance of traffic order and the development needs of the whole industry.The data transmission of VANET is mainly wireless transmission mode.This open environment makes vehicle nodes vulnerable to attack.The high-speed movement of network nodes and the change of network topology make it difficult to detect the attack behavior.Because of its particularity,the existing network security technology of VANET is not completely applicable.For V2 X communication,it challenges the information exchange of the outside world.At present,the research on information security of VANET mainly focuses on key management,identity authentication,message authentication and so on.These methods are the first layer of defense against known attacks.The VANET environment is open and complex.An effective means of detecting unknown attacks and suspicious behavior is needed.Anomaly detection technology can make up for the lack of encryption and authentication mechanism in VANET environment,and can be effectively supplemented for the VANET network defense.However,the traditional anomaly detection techniques are mainly oriented to computer networks.The vehicular ad hoc networks have the unique characteristics of high speed nodes,variable network topology,limited effective communication time between nodes and the unstable quality of wireless channel communication.Therefore,the traditional anomaly detection method is no longer applicable,a new anomaly detection method suitable for vehicular ad hoc networks must be studied.The problem of information security in vehicular ad hoc networks is studied in this paper.The characteristics and threats of VANET attacks are summarized,making an analysis of the challenges encountered in VANET information security.According to the characteristics of VANET intrusion detection model,the whole architecture is designed,and two kinds of VANET anomaly detection methods are proposed.The main research contents and work are as follows:(1)Summed up the causes of VANET information security problems and the development of domestic and overseas.(2)Starting from different angles,the information security threats faced by VANET network are analyzed.The characteristics of attack behavior in VANET are summarized and classifies according to their characteristics.This paper analyzes the threat of VANET attacks on the availability,reliability and other aspects of VANET networks.By analyzing the attack behavior and threats in VANET,we can find the security vulnerabilities in the VANET domain information security link.The main problems encountered in the implementation of the VANET security program are analyzed with analysis of possible layer by layer problems for the WAVE(Wireless Access in Vehicular Environments)architecture.The application of security technology is summarized.(3)The problems that the traditional intrusion detection technology may face in the application of VANET are analyzed,and made a formal description of intrusion and detection behavior in VANET.This paper summarizes the current methods used in intrusion detection technology,analyzes the existing research results of VANET intrusion detection technology,and summarizes the characteristics that VANET intrusion detection system should have.An overall architecture design of VANET intrusion detection system is proposed.The local detection mechanism of vehicle nodes and the cooperative response mechanism are introduced.(4)An entropy based VANET anomaly detection method is proposed.The method mainly detects the flooding attacks of security messages in VANET,and detects the Sybil attacks and replay attacks.The method makes a theoretical analysis of entropy changes of flooding and replay attacks,relative entropy changes of Sybil attacks.It is proved that VANET anomaly detection can be carried out by using entropy.The model of anomaly detection based on information entropy is established,and the detection process is divided into three parts.A VANET attack simulation system is established and verified by the above theoretical analysis.The experimental results show that the entropy based VANET anomaly detection method is suitable for VANET networks.(5)A method of VANET anomaly detection based on random forest is proposed.According to the characteristics of random forest model,a VANET anomaly detection framework based on random forest is proposed.The feature attributes are optimized and sorted.Using the optimized grid search method,the optimal parameter set is determined and the final model of random forest is generated.The random forest model is validated by the test data set,and the accuracy is higher.With the introduction of fresh data,the evaluation of data sets and the revalidation of the model,it has resulted in a decrease in accuracy.A random forest optimization method is proposed to further improve the accuracy of the model.The random forest model,the optimized random forest model and the decision tree model are compared.The results show that the decision tree model has the lowest detection accuracy,and the optimized stochastic forest model is superior to the other two models.