Severity-First Risk Prioritization for AI Poisoning in Edge-Integrated Internet of Vehicles
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The integration of artificial intelligence (AI) and edge computing in the Internet of vehicles (IoV) enables real-time perception and decision-making, but also introduces new cybersecurity risks. Among these, data poisoning attacks can compromise training pipelines, sensor inputs, and model updates, leading to hazardous system behavior. Traditional failure modes and effects analysis (FMEA) based on risk priority number may under-prioritize rare but severe adversarial failures. This paper presents a severity-first risk prioritization framework that applies action priority (AP) logic using structured severity-occurrence-detection (S-O-D) scoring. We map poisoning attack vectors to failure modes and operational consequences in edge-integrated IoV. We assign S-O-D scores adapted from AIAG-VDA FMEA Handbook, using expert judgment informed by standards and literature, and derive AP via rule-based, severity-dominant decision logic. Our findings show that most poisoning-related failure modes require high-priority action, emphasizing the need for severity-led prioritization in safety-critical edge-AI-enabled IoV. Mitigation measures are aligned with automotive cybersecurity standards and regulations for secure updates, vehicle-to-everything communication and edge-AI pipeline integrity, supporting practical deployment in vehicular edge environments.
Fields of Study & Subjects
Adversarial Robustness in Machine Learning
Safety Systems Engineering in Autonomy
Vehicular Ad Hoc Networks (VANETs)
Publication details
- Journal / source
- International Conference on Consumer Electronics
- Pages
- 1–6
- Publication date
- 3 February 2026
- Language
- EN