Periodic Reporting for period 1 - Affordable5G (High-tech and affordable 5G network roll-out to every corner)
Reporting period: 2020-09-01 to 2021-08-31
Affordable5G elaborated a system architecture that describes the components and building blocks in the private networks / NPN (non-public networks). Key enablers are i) disaggregated multi-vendor 5G RAN with open interfaces, ii) full-fledged 5G Standalone (SA) Core, iii) data analytics and AI/ML-based network optimization, iv) network slicing, iv) optimised hardware elements vi) RAN sharing and neutral hosting as cost-savings, vii) resource and service orchestration over heterogeneous infrastructures, and viii) efficient 5G front-haul network with support for TSN.
The RAN part follows the specifications provided by the O-RAN Alliance and includes components and open interfaces for the management of the O-RAN network functions and of the underlying cloud infrastructure, as well as for the incorporation of intelligent closed-loop automation targeting at cost-efficient network operation (zero-touch trend). The necessary resource and service orchestration in Affordable5G is undertaken by an end-to-end solution operating virtual network functions, MEC resources and hardware devices.
The outcome will be evaluated and validated in two vertical pilots in the second part of the project, namely the Mission Critical Service (MCS) and the Computer Vision Analytics for Emergencies (CVAE) services.
Affordable5G is a European project co-funded by the Horizon 2020 Programme of the European Commission.
• Adopting O-RAN Alliance specifications, Affordable5G RAN is implementing the option 7.2 low-layer split interface between RU and DU.
• Identification of the main 5G subsystems, its software components and categorization of hardware elements across the private network provision Leveraging ultra-low power accelerators for edge compute engines (running Machine Learning models for real-time inference at the edge, as well as supporting applications, running Deep Learning models).
• Implementing of novel transport and synchronization techniques and moreover address timing issues faced in the use cases (e.g. Time Sensitive Networks).
• Validation and selection Open-source solutions in many parts of networks.
• Implementation of smart control loops (near real and non-real time) with ML algorithms and integrated towards orchestration for RAN optimisation.
• Implementation and integration of the Slice Manager with the Orchestrator.
• Creation of new CU-UP instances per Public Land Mobile Network (PLMN) / Simple Network Management Protocol (SNMP) ID (for neutral hosting) and per network slice.
• Security assessment of the non-public networks has been reported, leveraging 3GPP relevant work.
• Several open-source frameworks relevant to Affordable5G concept have been reviewed and adopted in the project ecosystem, including O-RAN, TensorFlow Serving, Kubernetes, OSM, Prometheus, ArgoCD, etc.
• Deep assessment of the automation of the 5G network.
• Work on multiple complementary demos (6) covering different aspects of the network implementations and enhancements.
Affordable5G system requirements were elicited envisaged private network deployment scenarios and on three use cases regarding mission critical services, computer vision analytics for emergencies and manufacturing applications. Further elaboration on the system requirements and the architecture design, the thematic areas on which Affordable5G is introducing novelties are:
• O-RAN - The system will support disaggregated RAN with open interfaces following the O-RAN initiative to allow for flexible and cost-efficient 5G deployments in private and enterprise networks.
• 5GCore - To investigate the full potential of 5G for NPNs, the Affordable5G system architecture includes a full-fledged 5G SA core based on an extremely flexible software architecture, on top of which the 3GPP NFs are implemented. The core is capable of interworking with the project’s O-RAN-based radio access infrastructure.
• Data analytics and AI/ML-based network optimization - the system architecture incorporates components supporting both network and management data analytics services standardized by the 3GPP, as well as components for AI/ML-based O-RAN optimization.
• Slicing - The slicing solution specified in Affordable5G can provision end-to-end slices in the compute, network, and access network domains and perform slice resource partitioning at the infrastructure, network slice, and vertical service levels.
• Resource MANO – The orchestration specified for the Affordable5G system provides an end-to-end orchestration solution in 5G infrastructures, composed of heterogeneous components (virtual network functions, MEC resources and specific hardware devices) with various underlying virtualized infrastructure management technologies.
Affordable5G will impact society not only by providing support to the emergence of an European offer around disaggregated 5G technologies but also supporting several (9) SMEs in the consortium to consolidate their product competitive advantages and facilitate their commercialization paths.