Periodic Reporting for period 1 - PREDICT-6G (PRogrammable AI-Enabled DeterminIstiC neTworking for 6G)
Période du rapport: 2023-01-01 au 2023-12-31
PREDICT-6G’s mission is set towards the development of an end-to-end 6G solution including architecture and protocols that can guarantee seamless provisioning of services for vertical Use Cases (UC) requiring extremely tight timing and reliability constraints. To succeed, the solution will target determinism network infrastructures at large, including wired and wireless segments and their interconnections. PREDICT-6G will develop a novel multi-technology multi-domain data-plane (MDP) overhauling the reliability and time sensitiveness design features existing in current wired and wireless standards. The ambition is for the MDP design to be inherently deterministic. To achieve this, PREDICT-6G will develop an artificial intelligence -driven multi-stakeholder inter-domain control-plane (AICP) for the provisioning of deterministic network paths to support time sensitive services as requested by end-customers and with different scaling ambitions, e.g. from the network in a single vehicle to a large, geographically dispersed network. This requires timely monitoring and prediction of the behaviour of the complete network, including identifying potential sources of quality violations and analysing various routes of the traffic flows. These capabilities will be delivered through the PREDICT-6G artificial intelligence-powered digital twin framework, allowing the prediction of the behaviour of the end-to-end network infrastructure, and enabling anticipative control and validation of the network provisions to meet the real-world time-sensitive and reliability requirements of the running services.
The identified project objectives are listed below:
- O1: Define a deterministic (i.e. reliable, time sensitive, and predictable) multi-technology, multi-stakeholder network. Provide the PREDICT-6G network architecture based on the analysis of relevant UCs, evaluating their potential business impact.
- O2: Design the PREDICT-6G multi-technology, multi-domain data-plane.
- O3: Provide a novel AICP to enable determinism in multi-stakeholder multi-technology 6G environments.
- O4: Integration, experimentation, and UC verification.
- O5: Disseminate and communicate project results to the broadest audience using Open Science practices, impact standards and European Research and Innovation ecosystems, and exploit project outcomes
The identified project objectives are listed below:
- O1: Define a deterministic (i.e. reliable, time sensitive, and predictable) multi-technology, multi-stakeholder network. Provide the PREDICT-6G network architecture based on the analysis of relevant UCs, evaluating their potential business impact.
- O2: Design the PREDICT-6G multi-technology, multi-domain data-plane.
- O3: Provide a novel AICP to enable determinism in multi-stakeholder multi-technology 6G environments.
- O4: Integration, experimentation, and UC verification.
- O5: Disseminate and communicate project results to the broadest audience using Open Science practices, impact standards and European Research and Innovation ecosystems, and exploit project outcomes
During the first year, the project completed the use case definition, requirement analysis and detailed reference architecture specification of PREDICT-6G, resulting in a system definition that can be adapted to different network technologies, physical/virtual devices, HW/cloud/compute infrastructure, and lab environments. On the data plane, the project identified and specified technical enablers of creating cross-domain deterministic services – such as deterministic service continuity at domain borders, deterministic scheduler designs, cross-domain split of reliability and deterministic enablers, means of time sync across multiple domains. Programmability of the data plan enablers were framed in a modular and extendible OpenAPI framework wrapping technology specific deterministic capabilities into technology agnostic control/management plane services to enable the composition of e2e deterministic services spanning multiple domains and technologies. The project defined the information model for e2e deterministic services and the e2e-domain level service continuum, which allows the segmentation of e2e deterministic service requirements to domain level responsibilities by considering the capabilities and resources present in the constituent network domains.
Regarding its standardisation activities, PREDICT-6G impacted the international SDOs 3GPP, ETSI, IETF and O-RAN with 15 meaningful technical contributions, as reported in D5.3. The project also contributed to the pre-standardisation effort NGMN with two technical contributions. All standardisation work originated from the R&D efforts of one or more PREDICT-6G organisations and are directly related to the work conducted in the first period.
Regarding its standardisation activities, PREDICT-6G impacted the international SDOs 3GPP, ETSI, IETF and O-RAN with 15 meaningful technical contributions, as reported in D5.3. The project also contributed to the pre-standardisation effort NGMN with two technical contributions. All standardisation work originated from the R&D efforts of one or more PREDICT-6G organisations and are directly related to the work conducted in the first period.
System Architecture, Use cases and Business Impact Work Package (WP1), has focused on the architecture definition part and use cases analysis. During the first period, WP1 has produced the following key achievements: (1) analysis of use cases and identification of relevant requirements from them serving as input for the architectural work; (2) initial design of the PREDICT-6G architecture are reference input for the technical work packages dealing with control and data plane developments (documented in D1.2); (3) initial discussions on the security aspects of the project (some preliminary ideas reflected in D1.2).
The key results beyond the state of the art regarding the Multi-technology multi-domain Data-Plane (MDP, WP2) during the first year have been: (i) stablishing the collaboration between partners in the Layer 2 enhancements for determinism, (ii) creating the methodology for defining the OpenAPIs exposed towards AICP and (iii) defining the multi-domain architecture of the project. One of the key achievements is the setup of collaboration clusters in the research area, which has been prolific and we have 6 accepted scientific papers during the first year of the project (Please refer to D2.1 and D2.2). Regarding OpenAPIs, we established the methodology for defining them and during the first year we defined together with WP3 (AICP) the development cycles to be achieve during the second and third year of the project. Finally, we define a very ambitious architecture, based on DetNet, for scenarios with technology and/or administrative multi-domain, as detailed in D2.2
Regarding the work on the AI-driven Multi-stakeholder Inter-domain Control-Plane (AICP, WP3), during the first year the following beyond of the state of the art results have been achieved: i) Design of a complete functional architecture for the AICP with the definition of the minimum set of procedures to be supported for the management of an E2E deterministic service. The AICP functional architecture follows the SBA paradigm, specifying a set of services (Management Services, MS) in charge of providing all the AICP functionalities and ii) Basis of the abstraction and integration of different network technologies building the MDP. These results have been reported in D3.1 (“Release 1 of AI-driven inter-domain network control, management, and orchestration innovations”) at M9 as part of MS3 and disseminated with a scientific paper in 6G-PDN Workshop co-located with ACM MobiHoc ’23.
During the first reporting period, Integration and use case verification (WP4) have been working on the formalisation of an integration methodology to be followed during the execution of the project. This integration methodology leverages the interfaces, and workflows defined in WP1 and is based on the definition two different level of tests: (i) infrastructure integration test in which the aim is to validate the correct implementation of a specific of the generic interface by the MDP functionality; (ii) E2E UC validation tests and activities focus on the KPIs and functional validation of the AICP platform. For this, within this task we have developed specific templates to convey the information of the different tests at different levels.
Also, detailed activities for both data plane and control plane components are provided and 4 cycles of releases plus a 3-month contingency are anticipated.
The key results beyond the state of the art regarding the Multi-technology multi-domain Data-Plane (MDP, WP2) during the first year have been: (i) stablishing the collaboration between partners in the Layer 2 enhancements for determinism, (ii) creating the methodology for defining the OpenAPIs exposed towards AICP and (iii) defining the multi-domain architecture of the project. One of the key achievements is the setup of collaboration clusters in the research area, which has been prolific and we have 6 accepted scientific papers during the first year of the project (Please refer to D2.1 and D2.2). Regarding OpenAPIs, we established the methodology for defining them and during the first year we defined together with WP3 (AICP) the development cycles to be achieve during the second and third year of the project. Finally, we define a very ambitious architecture, based on DetNet, for scenarios with technology and/or administrative multi-domain, as detailed in D2.2
Regarding the work on the AI-driven Multi-stakeholder Inter-domain Control-Plane (AICP, WP3), during the first year the following beyond of the state of the art results have been achieved: i) Design of a complete functional architecture for the AICP with the definition of the minimum set of procedures to be supported for the management of an E2E deterministic service. The AICP functional architecture follows the SBA paradigm, specifying a set of services (Management Services, MS) in charge of providing all the AICP functionalities and ii) Basis of the abstraction and integration of different network technologies building the MDP. These results have been reported in D3.1 (“Release 1 of AI-driven inter-domain network control, management, and orchestration innovations”) at M9 as part of MS3 and disseminated with a scientific paper in 6G-PDN Workshop co-located with ACM MobiHoc ’23.
During the first reporting period, Integration and use case verification (WP4) have been working on the formalisation of an integration methodology to be followed during the execution of the project. This integration methodology leverages the interfaces, and workflows defined in WP1 and is based on the definition two different level of tests: (i) infrastructure integration test in which the aim is to validate the correct implementation of a specific of the generic interface by the MDP functionality; (ii) E2E UC validation tests and activities focus on the KPIs and functional validation of the AICP platform. For this, within this task we have developed specific templates to convey the information of the different tests at different levels.
Also, detailed activities for both data plane and control plane components are provided and 4 cycles of releases plus a 3-month contingency are anticipated.