The School of Computer Science is pleased to present…
Misbehavior Detection in Vehicular networks using Unsupervised Algorithms
MSc Thesis Defense by: Girish Nathi
Date: Thursday, 13 Jun 2024
Time: 3:00 PM
Location: Essex Hall, Room 122
Abstract: A wireless network connects moving or stationary automobiles and other infrastructure nodes to form a Vehicular Ad Hoc Network (VANET). One of the primary functions of VANETs is to ensure drivers' comfort and safety by sharing safety and non-safety information, including periodic broadcasts of Basic Safety Messages (BSM), which contain relevant status information such as position, heading, and speed. Inserting false information in BSMs can have serious consequences, such as traffic jams or accidents, making the detection of position falsification attacks crucial. A number of supervised learning algorithms have been proposed for detecting such attacks. This thesis addresses the challenge of detecting position falsification attacks using unsupervised machine learning models, given the difficulty in obtaining labeled datasets in real-world scenarios. We developed and evaluated these models using the publicly available Vehicular Reference Misbehavior (VeReMi) dataset, based on simulated traffic data. Our proposed models, especially the DBSCAN algorithm, showed results comparable to existing supervised techniques. These results highlight the potential of unsupervised learning in enhancing the security framework of VANETs.
Keywords: VANETs, Misbehavior detection, Position Falsification Attack, VeReMi Dataset, Machine Learning, Unsupervised Algorithms
Thesis Committee:
Internal Reader: Dr. Muhammad Asaduzzaman
External Reader: Dr. Ahmed Hamdi Sakr
Advisor: Dr. Arunita Jaekel
Chair: Dr. Jessica Chen
