MSc Thesis Proposal: Plausibility based Framework for Data Centric Misbehavior Detection in Vehicular Networks by Amitoj Birah

 

 

Amitoj Birah

 

Date: April 6th, 2026

Time: 10:00 am- 11:30 am

Location: 122 Essex Hall

 

Abstract: Vehicle Ad Hoc Networks (VANETs) rely on the timely exchange of safety messages to enable cooperative driving applications. Insider vehicles with genuine credentials, on the other hand, may broadcast inaccurate information, jeopardising system reliability and road safety. Existing data-centric misbehavior detection methods frequently rely on static thresholds, supervised learning, or cooperative infrastructures, limiting their adaptability and deployment feasibility in dynamic traffic situations. This research describes a local, data-centric misbehavior detection approach that combines physics-based plausibility checks and unsupervised anomaly identification. The suggested framework relies solely on locally received signals and is trained with benign data, removing the need for labeled attack samples or inter-vehicle cooperation. Unlike existing approaches that treat misbehavior detection as a supervised classification problem, this work reframes it as an unsupervised anomaly modeling task grounded in physical feasibility constraints.

 

Index Terms—VANET, misbehavior detection, plausibility check, BSM

 

Thesis Committee: Internal Reader: Dr Arunita Jaekel , Dr Alioune Ngom Advisor: Dr Ikjot Saini