Research On The Sybil Attack Detection In VANET Based On Geographic Location

VANET(Vehicular ad-hoc network)is a powerful platform for direct communication between vehicles.The platform can be used for traffic management,enhancing road safety and serving the passengers and drivers.With the continuous research and development of Vehicular ad-hoc network,the riding and driving experience of the passengers and drivers is better than before.But the joining of wireless network inevitably makes many security problems in traditional wireless network have the same or more terrible threat to the vehicular ad-hoc network.One of them,the name of which is the Sybil attack has been attracting increasing attention.The Sybil attack is embodied in the fact that a physical attack subject has multiple identifiers and the participation of multiple identities in the network weakens the role that honest nodes can play and ultimately achieves the purpose of the attack.Sybil attacks generally start attacks by creating multiple false identities and sending multiple false messages using different identities in order to disrupting and deceiving other nodes in the network.There is a certain correlation between the attribute characteristics of multiple false identities of Sybil attacks,such as the geographic location information,motion trajectory,time of receiving external messages,signal intensity detected when receiving messages from other nodes,etc..However,multiple identities may be assigned different parameter information in order to launch an attack,such as assigning different geographic location information for vehicles of different identities,so that other honest vehicles receive different corresponding information from the actual ones and achieve some degree of deception.In response to the characteristics of Sybil attacks,a witness selection strategy WCMR(Weakest correlation,minimum redundancy)is proposed,which uses the correlation between the witness node and the Sybil node or the redundancy between the candidate and the selected witness collection to select the node with the least correlation to the detected nodes or with the least redundancy to other nodes as the witness.Sybil nodes and malicious nodes are excluded from the pool of candidates to be selected as much as possible to obtain a more reliable set of witnesses.After the collection of witnesses,because of the certain degree of difference in the credibility of each witness,using the relevant method we can obtain the measure of the credibility of the node which is proportional to the credibility of the node.According to the metric value,the corresponding weight value is set for each vehicle node's testimony information.The greater the credibility of the witness node is,the larger the corresponding weight value is.In this way,different witness nodes play different roles in the detection process,which reduces the interference of hidden Sybil nodes or malicious nodes and improves the honest node in playing a positive impact in the detection process.When there is a Sybil node or a malicious node in the witness set,the method of setting the weight value can reduce the adverse effect on the detection result and increase the detection rate of the Sybil attack in the vehicle network.Based on the proposed scheme,this article uses wireless communication simulation software ns-3 and road simulation software sumo to do a large number of experiments using some observation variables such as observation time and detection deviation.It is concluded that this scheme can detect the Sybil nodes or malicious nodes with high detection rate,and exclude the Sybil nodes and malicious nodes from the normal vehicle communication environment.