ETD

• anomaly detection system
• automotive cybersecurity

The aim of this work is to design, validate and integrate into hardware an anomaly-based IDS. After a brief introduction to the problem and an overview on the state of art, it explores the CANini dataset, derived from the CAN-MIRGU repository, which serves as the foundation for the development of the system. A preprocessing phase was conducted to adapt the dataset to simulation eviroment, followed by the analysis of the features, to select those which were considered significant.
The proposed anomaly-based, single layer IDS was first modeled and simulated using Matlab, where its performance was validated through key metrics (e.g. accuracy). The system was then embedded into the ShieldBuddy TC375, an automotive-grade board, to demonstrate its feasibility in a real-time hardware environment. The results confirm that the developed IDS can effectively distinguish between legitimate traffic and malicious injections, providing a robust layer of defense for vehicular networks.