Applying Machine Learning to Enhance Security in Vehicular Ad hoc Network

Abstract

Road safety remains a challenging task for government and researchers, as the newlinenumber of vehicles on the road is increasing and the road space is limited which leads newlineto increasing the density of vehicles running on the road. Proper management is the newlineonly solution that remains. In order to manage the vehicles many researchers have put newlineforth their research, Intelligent transportation system is one of the most prominent newlinesolutions to enhance road safety and reliable transportation. Intelligent transportation newlinecan be achieved through the use of vehicular ad hoc networks, which is one of the newlinefields. This comprehensive communication is the combination of a decentralized ad newlinehoc network with a centralized client-server architecture. Vehicles running on the newlineroad are using IEEE 802.11p dedicated short range technology to connect with each newlineother and devices installed on the roadside. These devices that are installed on the newlineroadside form an infrastructure and connect to the internet through a client-server newlinearchitecture. newlineThe dynamic nature of VANET makes it a challenging task to deploy in real-world newlinescenarios. The network is traversed by numerous vehicles. The on-board unit that is newlineembedded in vehicles has limited memory space, limited energy, and requires fast newlinecommunication. Despite these limitations, it is challenging to secure a highly dynamic newlinenetwork from various security attacks. To achieve secure vehicular communication, newlinevarious cryptographic methods are used to authenticate the vehicles and other nodes. newlineIt is necessary to make extra efforts to identify attackers who have valid credentials newlineand surpass cryptographic methods. Machine learning/deep learning methodologies newlinehave the potential to be extremely useful in identifying these attackers and preventing newlinethem. The limited capacity of the on-board unit will result in slower communication if newlinea machine learning model is deployed. The efficiency of communication will not be newlinesignificantly affected if these models are deployed in roadside installed devices and newlinethe results are re