We are pleased to present the outcomes of our team’s participation in Threat Modeling Hackathon 2025, where we focused on developing a comprehensive threat model for an Autonomous Electric Car, with a specific deep dive into the Autonomous Driving Software Stack based on ROS 2 architecture.
Project Scope and Approach
Our primary objective was to model threats for the autonomous driving capabilities of the Autonomous Car. We began by constructing a C4 model to outline the high-level software architecture of the vehicle.
Following initial research and iterative discussions, we refined our scope to focus specifically on the ROS 2 system architecture — detailing Nodes, Namespaces, Topics, Publishers, and Subscribers that form the communication fabric of autonomous driving functions.
After evaluating different industry architectures (notably comparing AUTOSAR and ROS 2), we selected ROS 2 for its modular, publisher-subscriber model that allowed better transparency and traceability for threat modeling activities.
To identify and document threats systematically, we:
- Mapped connectivity and data flow across all critical modules (Perception, Planning, Control, V2X Communication).
- Evaluated multiple open-source and community threat modeling tools (IriusRisk, ThreatCanvas, Microsoft TMT), and ultimately developed a customized Excel-based threat model template tailored to our needs.
- Chose to apply the TARA (Threat Analysis and Risk Assessment) methodology, based on ISO/SAE 21434 principles, due to its structured evaluation of attack feasibility, impact rating, and calculation of residual risk.
Deliverables
We are proud to share the following deliverables as part of our project output. The folder includes the:
- Threat Model
- Architecture Diagram
- Video Presentation
- Retrospective Document
