Enhancing Open RAN Security with Zero Trust and Machine Learning
Hajar Moudoud, Soumaya Cherkaoui
Abstract
As 5G networks continue to evolve, they are becoming increasingly intricate and diverse, accommodating a vast array of devices. This complexity poses significant challenges when it comes to safeguarding these networks against cyber-attacks. While the core infrastructure of 5G is shifting towards virtualization and is being deployed by multiple vendors, Radio Access Networks (RANs) have traditionally been delivered as tightly integrated solutions, often lacking interoperability. Open RAN (O- RAN) emerges as a flexible and cost-effective approach to designing and deploying mobile networks. It allows for the integration of mobile radio access networks from various vendors through the use of disaggregated and O- RAN technologies. Nevertheless, the introduction of components from multiple vendors into the supply chain increases complexity, making it difficult to ensure the security of each individual component. Additionally, O- RAN's attack surface expands due to seamless access for numerous devices. In this dynamic landscape, adopting a zero-trust architecture (ZTA) presents an attractive framework for bolstering security in open networks. We introduce an intelligent architectural concept design that leverages key zero-trust principles to enhance information security within the inherently untrusted O-RAN environment. Moreover, we propose a solution that combines deep reinforcement learning techniques with traditional machine learning methods to fortify security in Open RAN. Finally, we evaluate the performance of our proposed solution using the UNSW network dataset and demonstrate its superior performance across selected metrics.