6G-Machine Intelligence based Radio Access Infrastructure (6G-MIRAI) aims at developing reliable and robust AI-native wireless communication systems that enable the practical exploitation of the full potential of the latest physical layer technological advances, especially cell-free massive MIMO, and of next-generation virtualized and potentially disaggregated radio access networks (RANs). To achieve this goal, 6G-MIRAI will rely on extensive know-how from established industry players and academic partners from both the EU and Japan. The overall goal of 6G-MIRAI is deconstructed into four main objectives: (O1) Reliable and robust AI-ML techniques for future wireless communications; (O2) Practical AI-native design of next-generation radio access networks; (O3) Common EU-JP platform for data, benchmarking, and validation; and (O4) Aligned EU-JP strategy on future standardization efforts. These objectives align with the EU HORIZON-JU-SNS-2024-STREAM-B-01-05 call requirements, focusing on the evolution of Radio Access Networks (RAN) for 6G to pave the way for future advancements towards AI-native radio access networks. All four objectives will contribute to aligning views on the radio interface and RAN concepts between the EU and Japan, and between main industry and academic institutions in the EU. 6G-MIRAI objectives will be achieved through five interconnected research and technology items: (RTI1) Realistic channel and hardware models to enable the AI-native air interface; (RTI2) Practical AI-based physical layer designs; (RTI3) Scalable AI-based network coordination; (RTI4) Evolved AI-ready architecture designs; and (RTI5) AI-oriented data management, testing, and proof-of-concept. These RTIs will enrich the technological portfolio of all involved partners and the 6G ecosystem, since they will have a direct impact on standardization, IP generation, and scientific publications.

NEWTON proposes a revolutionary approach of a converged wireless-optical network architecture that makes reality an efficient 6G Cell-Free (CF)-based access network for high-density and high-coverage deployments. The proposed architecture takes also full advantage of novel solutions for both the radio-edge and regional-edge domains, as well as for efficient network-management approaches, targeting to achieve sub-second latency, more than million devices per square kilometer connection density and user-experienced data-rate up to 10Gbps. NEWTON research program will engage in cutting-edge research to advance and assess a wide range of techniques, tools and methodologies that will harness the innovative capabilities provided by the CF technology to their fullest extent and establish the groundwork for efficient end-to-end services in 6G network environments. The NEWTON approach will be directed toward pursuing three ambitious objectives, each aligned with an interdisciplinary but interconnected research domain. The three objectives are linked to challenges at the radio-edge, the regional-edge, and the network-management domains. At the radio-edge domain, NEWTON aims at developing Artificial-Intelligence (AI)-driven models in the access domain, focusing on the CF technology, targeting the efficient acquisition of the Channel State Information (CSI) in dense CF networks, while supporting infrastructures with high-number of antennas. In the regional-edge domain, NEWTON targets to develop scalable solutions for the interconnection of data-centers, to enable distributed neuromorphic architectures, while also being in line with the Open RAN initiative. Finally, at the network management domain, NEWTON will develop novel AI-based resource allocation solutions, targeting the optimal utilization of both networking and computational resources.

EMPOWER-6G proposes a novel converged optical-wireless architecture that realizes an efficient 6G Cell-Free (CF)-based access network for high-density and high-coverage deployments. The proposed architecture takes full advantage of the distributed processing CF concept, as well as of wireless mmWave solutions, while being in-line with the O-RAN Alliance. Specifically, the EMPOWER-6G vision is based on the development of novel solutions at the radio access domain by enabling emerging CF technologies, while also contributing innovations at the optical transport domain and significantly evolving the MEC system towards fully elastic Edge Computing. The EMPOWER-6G network configuration deploys a distributed Edge infrastructure with Data Centres (DCs) structured in 2 tiers, featuring Radio Edge Regional Edge nodes, where the former DCs hosts the Network Functions of the (virtualized) RAN fully aligned with the O-RAN specifications, while the latter DCs host non-real-time network functions. Thus, EMPOWER-6G will shape a novel 6G network configuration by addressing challenges at the radio-edge, the regional-edge, and the network-management domains. At the radio-edge domain, EMPOWER-6G aims at designing novel CF networking mechanisms that will allow the significant scaling up of Radio Unit (RU) deployment in a cost-effective manner, by exploiting the application of the distributed processing CF concept, based on the disaggregation of the traditional CF Central Processing Unit (CPU) into Distributed Units (DUs) and a Central Unit (CU), in line with the 3GPP NG-RAN architecture. In parallel, EMPOWER-6G proposes an innovative mmWave Hybrid MIMO solution, with beam-steering and beam-sharing support. At the regional-edge domain, EMPOWER-6G aims at providing a new design for an optical regional edge network with FTTH support in order to achieve optimum functional splitting options, while also supporting novel control-plane protocols for CF networking support.

The ambition of the NATWORK project is to set the foundations and deploy the very first economically realistic, energy efficient and viable bio-inspired AI-based 6G cybersecurity and resilience framework for intelligent networking and services, taking a holistic approach and considering all elements in a cross-sector business environment to address the diverse requirements and challenges that arise. The NATWORK project aims to develop a novel AI-leveraged self-adaptive security mechanism for 6G networks based on resilient bio-mimicry principles. The goal is to improve the malleability and the self-resilience of future 6G network ecosystems to offer augmented and secure services at the lowest energy costs. The principle premise is to empower various entities of 6G ecosystems with the ability to self-regulate their conditions to provide service continuity in compliance with service SLAs. The Secure Federated Learning architecture of NATWORK will be based on decentralized defensive AI models embedded in dis-aggregated 6G network physical layer, smart Edge Network Interface Cards and RAN devices with P4-based programmable data plane and advanced DPU acceleration, with local feature extraction at wire-speed and AI model training. Among the key 6G security challenges that NATWORK aims to alleviate are Moving Target Defense and adaptive response to incidents, the employment of Net Zero AI and energy-efficient security for sustainable networks, the Detection of new forms of attacks bearing deep control flow monitoring as well as the elaboration of a continuum of security for payload deployment fostering secure migration of novel forms of in-network operations and secure distributed computations.

EWOC project aims at developing a novel converged optical wireless network solution relying on a flexible, virtualizable infrastructure, required for full resource optimisation beyond 5G (B5G) requirements. Fundamental innovation will be sought through merging of the enabling concepts of optical layer virtualization, high frequency mm-wave transmission, multiple antenna technology, cell densification, terra-over-fiber (ToF) based femtocell connectivity and cloud radio access network (C- RAN) architecture. EWOC will aim at high capacity, low latency communications (40-90 GHz frequency), providing the basis for a 50-fold improvement over the 5G baseline. This necessitates development of novel, femto-cell technology, and seamless coexistence with first round legacy deployment. Such scenario also requires novel channel models and simulation methodologies to attain the desired trade-off between coverage, throughput and densification limits. EWOC will rely on fiber-optic deployment towards ToF connectivity, as an “added on feature” for the C-RAN architecture supporting resource management of versatile services with varying demands. Scenario compliant optical fronthaul virtualisation techniques, designed to provide cost effective beyond state-of-the-art resource optimisation, will be pursued through novel optical transceiver schemes and software defined network-based digital signal processing techniques. Research and training disciplines will serve as building blocks towards the scientific and socio-economic goals of increased capacity, coverage, flexibility, spectral efficiency, cost effectiveness, vendor agnosticism, and upgradability. EWOC provides a framework for promotion of such interdisciplinary innovation, with strong interoperability of models and methodologies from different disciplines. As such, EWOC training network is designed to foster opportunities for scientific and professional growth of ESRs from both topical and inter-disciplinary standpoints.