TELECOM LAB PROJECT
Energy-efficiency and security in O-RAN
This project is dedicated to redefining telecommunications for B5G/6G networks by addressing critical industry challenges such as cybersecurity, sustainability, and scalability, and by promoting interoperability and multi-vendor integration.
With security as a key enabler, it integrates advanced frameworks to mitigate risks like data poisoning, adversarial attacks, and interface breaches. Real-time anomaly detection, secure-by-design AI/ML models, and robust encryption ensure the integrity of critical services.
Simultaneously, the project prioritizes sustainability through energy-saving mechanisms and Digital Twin simulations, enabling efficient resource management and reducing environmental impact.
By combining cutting-edge technologies and strong collaboration with industry leaders, this project delivers future-ready solutions for the evolving demands of global connectivity.
USE CASE
The O-RAN Alliance places a strong emphasis on security, with Working Group 11 (WG11) leading efforts to develop security-focused specifications for secure-by-design O-RAN architecture.
This involves identifying potential security threats, assessing risks, defining security requirements, and establishing protocols specific to O-RAN components. WG11 also creates detailed test specifications to ensure these security measures are robust.
A key milestone in these efforts is the O-RAN Security Threat Modeling and Risk Assessment report, which defines critical assets, outlines threat models, and provides security principles alongside comprehensive risk assessments.
One of these threat models is the data poisoning of the RICs where an operator can use ML models trained with poisoned data and make real-time decisions dangerous for operating the network.
Effective security techniques to defend against data poisoning have been introduced such as robust training algorithms, AI/ML modified architectures and proactive techniques.
While security is a critical focus, it raises an important question: how can O-RAN balance the trade-off between robust security measures and energy efficiency? We explore the relationship between the security techniques for data poisoning and the energy costs associated with implementing these measures.
INNOVATION DIRECTION
Our innovation in system-level security focuses on:
Studying data poisoning attacks.
Defense mechanisms
Associated energy consumption in AI-driven systems.
Data poisoning, where malicious actors manipulate training data to compromise model integrity, poses a significant threat to modern networks. Through a comprehensive analysis of various defense techniques, we aim to understand their effectiveness against such attacks while quantifying their energy requirements.
By evaluating the trade-offs between security robustness and energy efficiency, our work seeks to identify optimal strategies that balance these critical factors. This approach not only enhances the resilience of AI systems but also ensures their sustainability, providing actionable insights for deploying secure and energy-efficient solutions in real-world scenarios.