Date of Award

5-28-2025

Publication Type

Thesis

Degree Name

M.Sc.

Department

Computer Science

Keywords

Collective Perception Messages; Collective Perception Services; Cooperative Intelligent Transport Systems; Vehicle-to-Everything; Machine learning

Supervisor

Arunita Jaekel

Rights

info:eu-repo/semantics/openAccess

Abstract

Cooperative Intelligent Transport Systems (C-ITS) are changing the way the vehicles are made to communicate by exchanging data in real time through the Vehicle-to-Everything (V2X) communication. However, with the increasing connectivity comes the risk of more threats. An adversary can launch various attacks by injecting false position and speed information to the system which may result in wrong traffic information, navigation, and even in an accident. Cooperative Perception Services (CPS) is one of the main components of C-ITS, where vehicles collaborate to provide each other with a richer understanding of their environment. But for these systems to be able to operate safely and properly, they have to be designed to counter misbehavior and other security risks. This thesis proposes a machine learning based method for detecting misbehavior in C-ITS using the SimCPS dataset. The dataset contains six attack types that target position and speed information across different attacker densities and target strategies. The research introduces a detection framework which analyzes collective perception messages to achieve effective malicious activity detection. The model learns attack patterns from the data itself instead of using predefined rules which enables it to handle various scenarios. The results show excellent detection performance, particularly in identifying subtle manipulations that traditional methods often overlook.

Download

Share

COinS