An Artificial Immune System Based Vehicular Misbehavior Detection System For Vehicular Ad-hoc Networks

Based on the powerful communication capability of Vehicular ad-hoc networks(VANETs),disruptive intruders may utilize VANET to launch their malicious attacks and cause different degree of damage to the normal order of the road traffic.Compared with mobile ad-hoc networks(MANETs),vehicular nodes in VANET have the characteristic of high mobility,their geographical distribution are usually wide.The security of VANET is closely linked with the safety of personnel and property,which puts forward higher requirements for both the reliability of network information transmission and the veracity of data processing in real-time.At present,the misbehavior detection methods for VANET are lacking of a macroscope analytical perspective in large-scale and distributed VANET environment.Besides,the detection efficiency is very low,which is far from meeting the safety requirements of the vehicular ad-hoc network.Based on the understanding of the artificial immune system(AIS)and the VANETs,this paper compares their similarities both in system characteristics and security aspects.The immune system protects the body from foreign harmful substances and remains robust in a variable environment,which meets the security requirement of the VANETs.This article imitating the immune mechanism in the immune system and dividing the misbehaviors in VANETs,then proposed a negative selection algorithm(NSA)based vehicular misbehavior detection method for VANETs.Considering so many features of the vehicles in VANETs,this paper uses the real value vector and variable-length detectors to improve coverage of detection.In order to simplify the computational complexity,the paper use the Euclidean distance to calculate the "affinity" of the detectors.After this,the clonal selection algorithm(CSA)is applied to improve the detection efficiency.In order to further improve the accuracy of the detection,this paper proposes a dendritic cell algorithm(DCA)detection system for vehicular misbehavior.This is a dynamic,parallel and distributed detection systems.Considered information gathering of vehicular nodes,misbehavior detection,and the follow-up tracking stages,this system can be used to classify various kinds of misbehaviors and realize the efficient global detection in VANETs.In the end,a simulation experiment based on the traffic simulator Vanetmobisim and the network simulator NS-2 proves that both the NSA-based detection system and the DCA-based detection system have high accuracy.The DCA-based detection system has a lower false positive rate.Both of them are well fitted to the features of VANETs.To the best of our knowledge,our artificial immune system based misbehavior detection technology is the first bionic computational intelligence(CI)detection system for VANETs.