Application Research Of Intrusion Detection System In Vehicle Networking

Vehicular ad hoc network(VANET)are considered to be the next important technology that will change our lives remarkably.The technology can make our lives and roads safer through allowing communication between vehicles.The study of its security is very important because VANET manages vital traffic information related human safety,security in VANET.To resist the security threat,intrusion detection system(IDS)is always deployed at each vehicle in VANET.IDS allows for detection of a suspicious activity within the VANET by analyzing messages from vehicles and triggering an alarm when these vehicles exhibits a malicious behavior.This paper has studied IDS in VANET.First,the characteristics and development of the IDS in VANET are introduced.Then we have analyzed advantages and disadvantages of the existing IDSs.Next,a novel IDS is proposed,which can be applied to VANET.Our main contributions are as following.1.We proposed GHSOM-based IDS for anomaly detection,which mainly consist of a novel feature extracting mechanism and GHSOM-based classifier.The proposed feature extracting mechanism is used to extract a vector with two features i.e.,traffic flow and vehicle position features.For efficiently extracting vehicle position feature,the extracting mechanism uses a semi-cooperative method to calculate the vehicle position feature through both position information of neighboring vehicles and position information of previous vehicles.The GHSOM-based classifier with two extra mechanisms is used to outlier detection,in order to detect deviation from vehicle's messages more accurately.In our experiment,we evaluate our system and observe that the proposed IDS is better than other available IDS in its measurement of accuracy,stability,processing efficiency and message scales.2.A novel filter model based on Markov methodology is proposed in IDS to reduce overheads and detection time without impairing the accuracy.This filter model forecast the future behavior(normal or abnormal)of neighboring vehicles to quickly filter messages from these vehicles instead of to detect these messages by detection mechanism.The filter model consist of three modules,i.e.,schedule,filter and update modules.The schedule module uses Baum–Welch algorithm to produce observation chain,which is used to forecast the future behavior in the filter module.According to the results of forecast,the update module use a timeliness method to update the observation chain for adapting change of behavior pattern.Experiments show that the proposed filter model can improve the detection time and overhead without affecting the detection accuracy.