5G Networks Through an Adversarial ML Lens
Overview
A research study on the adversarial risks of using AI/ML for 5G network automation. The work covers supervised, unsupervised, and reinforcement-learning attack surfaces through three case studies, proposes mitigation approaches, and offers guidelines for testing how well ML models hold up to attacks in 5G contexts. Published in IEEE Internet Computing 2021.
Outcome
Peer-reviewed publication that provides the 5G research community with a structured view of adversarial ML risks and mitigation strategies.
Screens & Demos
Architecture & Pipeline
-
01
5G ML Models
Supervised · Unsupervised · RL
-
02
Adversarial Attack
Custom Python envs
- Python
-
03
Robustness Evaluation
Three case studies
-
04
Mitigation Guidelines
Best practices
-
05
Published Findings
IEEE Internet Computing 2021
End-to-end flow, in order. Scroll sideways on a narrow screen.
Key Features
- Three case studies covering supervised, unsupervised, and RL attacks
- Custom Python environments for each adversarial scenario
- Mitigation guidelines and a framework for evaluating attack resistance
- Peer-reviewed publication in IEEE Internet Computing