Towards Machine-Learning-Based 5G and Beyond Intelligent Networks: The MARSAL Project Vision

Abstract

5G mobile networks will be soon available to handle all types of applications and to provide service to massive numbers of users. In this complex and dynamic network ecosystem, end-to-end performance analysis and optimization will be key features to effectively manage the diverse requirements imposed by multiple vertical industries over the same shared infrastructure. To enable such a vision, the MARSAL project targets the development and evaluation of a complete framework for the management and orchestration of network resources in 5G and beyond, by utilizing a converged optical-wireless network infrastructure in the access and fronthaul/midhaul segments. At the network design domain, MARSAL targets the development of novel cell-free based solutions by exploiting the application of the distributed cell-free concept and of the serial fronthaul approach, while contributing innovative functionalities to the O-RAN project. In parallel, in the fronthaul/midhaul segments MARSAL aims to radically increase the flexibility of optical access architectures via different levels of fixedmobile convergence. At the network and service management domain, MARSALwill exploit novel ML-based algorithms of both edge and midhaul DCs, by incorporating the Virtual Elastic DataCenters/Infrastructures paradigm. Finally, at the network security domain, MARSAL aims to introduce mechanisms that provide privacy and security to application workload and data, targeting to allow applications and users to maintain control over their data, while AI and and Blockchain technologies will be developed in order to guarantee a secured multi-tenant slicing environment.

Open or not open: Are conventional radio access networks more secure and trustworthy than Open-RAN?

“…In addition to the architectural disaggregation and openness (in the sense of interoperability), the aspects of cloudification and virtualisation [1], network slicing [2], [3] and machine learning [4] also play an important role in the Open RAN context. Yet, it is important to emphasize that except for disaggregation and openness, the other aspects such as virtualization and machine learning are not an inherent part of the Open RAN concept.…”

Cloud-native 5G experimental platform with over-the-air transmissions and end-to-end monitoring

“…Nevertheless, our experimental framework is designed to also work with other open-source 5G cores under acceptable adjustments. 2 B. Cloud-native deployment of the 5G core 5G is expected to support use cases that go beyond raw throughput performance, where the focus is to be put on service flexibility and agility. In this regard, the procedure to deploy 5G NFs has a critical impact.…”

“…In turn, this enables multi-access edge computing (MEC) applications for low-latency and high bandwidth like virtual and augmented reality (VR/AR). In this context, projects like MARSAL [1], [2] propose a new paradigm of elastic virtual infrastructures that integrate transparently a variety of novel radio access, networking, management, and security technologies to deliver end-to-end transfer, processing, and storage services in an efficient and secured way.…”

A Survey on Research Activities for Deploying Cell Free Massive MIMO towards Beyond 5G

“…In [48], an implementation of the cell free massive MIMO based on the cloud baseband architecture is introduced. It is expected that the cell free massive is operated flexibly for satisfying the diversified communication requirements by utilizing the degrees of freedom of the large number of distributed antennas and the computing resource efficiently with such the platform [49], [50].…”