The increasing number of vehicles on the roads over the past few decades is leading to more problems, including collisions, congestion, air pollution and fuel consumption, for traffic management authorities, drivers, and others. According to a report published by Canadian Motor Vehicle Traffic Collision in 2018, there were around 150,000 injuries, 11.5 million hours spent on the road, and 22 million liters of fuel drained annually in Canada only. With the emergence of new technologies aimed at promoting smart cities and Intelligent Transportation Systems (ITS), addressing above mentioned issues are critical for road traffic and vehicles. Vehicular Ad hoc Network (VANET), a sub-class of Mobile Ad hoc Network (MANET), allows safety and non-safety communications between vehicles and infrastructure nodes, and is a critical enabling technology for future ITS. In vehicular ad hoc networks (VANETs), vehicles exchange sensitive information, including vehicle identity, position, speed, heading and other parameters, with nearby vehicles or infrastructure through Basic Safety Messages (BSM). However, VANET communication is vulnerable to various types of attacks, and appropriate security mechanisms must be in place to ensure that exchanged messages are not altered or false messages created by malicious attackers. VANET communication should ensure security in terms of availability, confidentiality, authentication, and integrity because compromised communication can pose a serious threat to the safety and well-being of users. In V2V communication, the exchanged information (emergency messages, safety messages, etc.) through wireless channels requires a secure environment to avoid attacks on the V2V network. Such attacks include position forgery, Sybil, Replay, DoS, etc.

In recent years, traditional ML-based approaches have achieved promising results for certain types of attacks, e.g., position falsification attacks. However, other types of attacks, e.g., Sybil attacks, are more difficult to detect. ML techniques have shown great potential for fast, real-time detection of different types of attacks in VANET. However, one inherent challenge in this area is to obtain realistic, large-scale datasets for vehicular networks. Transfer learning based models have been able to effectively address this issue in different areas while providing improved performance and requiring fewer resources. The primary objective of this research is to develop an Intrusion Detection System (IDS) that is specifically engineered for Basic Safety Message (BSM) data in connected vehicles using innovative Transfer Learning approaches. We propose Deep Transfer Learning based IDS (DTF-IDS) using structured and unstructured approaches for BSM data. The proposed techniques aim to enhance detection accuracy, reduce training time, and detect attacks under different traffic scenarios with low and high density traffic.

 

Internal Reader: Dr. Imran Ahmad

Internal Reader: Dr. Boubakeur Boufama             

External Reader: Dr. Ning Zhang

Advisor(s): Dr. Arunita Jaekel