By Antonio de la Oliva, Project Coordinator
The end of 2023 marks also the closing of the first year of the PREDICT-6G project. During this year, several key milestones have been achieved, setting the basis for the technical work of the second year. During the first months of 2023, PREDICT-6G focused on understanding the use cases which may benefit from deterministic networking innovations. The three use cases defined in the proposal – Smart manufacturing, Deterministic services for critical communications and multi-domain deterministic communications – have been extended to Localisation and Sensing, XR and IETF RAW and DetNet. With the inclusion of these new use cases, we believe the project sufficiently covers the requirements of current industries.
Based on the requirements and expected KPIs of the identified use cases, the main efforts of the year 2023 were dedicated to developing the system architecture and the principles of the Multi-Domain Dataplane (MDP) in WP2 and the AI-driven multi-stakeholder inter-domain Control Plane (AICP) in WP3. The work on the MDP and AICP design was synthetised into a full architecture specification, including E2E and domain level management service definition, E2E deterministic service architecture and lifecycle, MDP architecture and cross-domain U-plane integration enablers, system and service procedures, and Open APIs for technology integration with 3GPP and IETF DetNet domains.
Regarding the MDP, innovation areas were explored in the data-plane to enable cross-domain deterministic services, such as deterministic service continuity at domain borders, deterministic scheduler designs, cross-domain split of reliability and deterministic enablers, and time sync across multiple domains. Technologies consider 3GPP (with and without IETF DetNet integration support), IEEE TSC (with IETF DetNet), and novel Wi-Fi MAC capabilities. To support these technologies at the control level, the project follows an OpenAPI methodology defining the programmability layer on top of a network segment with the selected deterministic technologies, or even over network domains with no explicit support for determinism. The OpenAPIs are the interface between the PREDICT-6G MDP and AICP, enabling the technology-agnostic design for the AICP architecture. Current discussion within the project is the relation of the OpenAPIs with the DetNet component, and whether all different domains should be integrated below the DetNet control.
At its own pace, the AICP has also evolved significantly during 2023. The AICP follows a service-based architecture principle, where Management Services (MS) interact with each other via clearly defined APIs. MSs that share the same management scope are organised into Management Domains (MD). AICP defined a single MD for the E2E deterministic services, as well as separate MDs for each of the technology domains that are part of the initial release of PREDICT-6G’s MDP. This role split enables the AICP to govern any number of technological domains (i.e., 3GPP, IETF DetNet / IEEE TSN, Wi-Fi, and more) driven by the same end-to-end service focus. Individual Management Services implement capabilities such as service ingestion, service automation, path computation, digital twinning, AI-based predictions, measurement collection, and the exposure of topology, services, resources, and capabilities from domain level to end-to-end and from end-to-end to the consumer (e.g., user, operator) of PREDICT-6G.
At the end of 2023, the project also started the development and integration activities. Led by WP4, the first step was to define the methodology (sprint based agile development and integration cycles) and timeline for developing, integrating, and validating the PREDICT-6G system components. The design of Open Labs and the structure of the validation tests have also been central in this first year.
In conclusion, although the project is still in its infancy, we have a clear view on its merits and innovations, and the consortium is focused on achieving great impact in the industry, developing new technologies to create new deterministic networks.