Using deep learning to address the security issue in intelligent transportation systems

Abstract

The lives of people are at risk from security and safety risks with Intelligent Transportation Systems (ITS), particularly Autonomous Vehicles. In contrast to manual vehicles, the Security of an AV’s computer and communications components may be penetrated using sophisticated hacking methods, preventing us from employing AVs in our daily lives. The Internet of Vehicles, which connects manual automobiles to the Internet, is vulnerable to cyber-attacks such as lack of service, spoofing, sniffer, widespread denial of service and repeat attacks. This paper presents a unique intrusion detection system for ITS, using Enhanced Cuttle Fish Optimized Multiscale Convolution Neural Network (ECFO-MCNN), that uses vehicles to identify networks and infrastructure and detects careful network activity of in-vehicle networks. The primary goal of the suggested strategy is to identify forward events emanating through AVs’ central network gateways. Two benchmark datasets, namely the UNSWNB15 dataset for external network communications and the car hacking dataset for in-vehicle communications, are used to assess the proposed IDS. The evaluation’s findings showed that the performance of our suggested system is superior to that of traditional intrusion detection methods.