A Learning-Based Zero-Trust Architecture for 6G and Future Networks

Abstract

In the evolution of 6G and Future Networks, a dynamic, flexible, learning-based security architecture will be essential with the ability to handle both current and evolving cybersecurity threats. This is specially critical with future networks' increased reliance on distributed learning-based approaches for operation. To address this challenge, a distributed learning framework must provide security and trust in an integrated fashion. In contrast to existing approach such as federated learning (FL), that update parameters of a shared model, this work proposes an architecture that is capable of integrating advanced learning with real-time digital forensics, e.g. monitoring compute and storage resources. With real-time monitoring, it is possible to develop a learning-based, real-time Zero-Trust Architecture (ZTA) to achieve the high levels of security. The proposed architecture, serves as a framework to enable and spur innovation, where new machine learning based techniques can be developed for enhanced real-time, adaptive and proactive security, thus, embedding future networks' security with learning-based ZTA elements.