Enhanced security mechanism in vehicular networks using ensemble machine learning to detect malicious activity in VANETs

One Vehicular Ad-Hoc Network (VANET) consists of numerous active and static vehicles connected in a wireless network. It is an easy and affordable method for transferring information related to traffic and vehicles to the traffic control rooms but may cause the security issues. VANET uses certain protocols for securely transmitting the information and providing internet connectivity to vehicle nodes. An AODV is commonly used in VANET. It is a fast and low-processing machine language model with memory overhead. Vehicles are equipped with OBU, which executes transferring message. If numerous vehicles send their information related to the vehicle simultaneously, then the information gets lost due to collision. Yet, the functions of nodes change frequently, which causes difficulty in routing due to software or hardware default in vehicle. To overcome the abovementioned issues, detection of new attack in VANET is designed. Initially, the data are gathered from online sources. To eliminate unnecessary features from data, data cleaning is performed. To organize the data, the normalization technique is applied. From this data, optimal weighted features are selected using Modernized Random Parameter-based Green Anaconda Optimization (MRPGAO). The attack detection is performed using the developed Ensemble Machine Learning Model (EMLM) for providing security in VANET network. It is formed by combining the Multi-Layer Perception, SVM, AdaBoost and Bayesian network. Then the fuzzy ranking method is employed to classify the types of attack in VANET. Finally, the performance of the developed secure machine learning-based attack detection in VANET is validated by comparing it with existing techniques and algorithms.