System Architecture

The scope covers the realisation of a unified and open communication and computing architecture beyond the current 5G SBA capabilities. Such architecture will enable seamless operations across a multiplicity of heterogeneous domains, infrastructures, services, business, and application heterogeneous domains, paving the way towards massive digitisation. The architecture should be agnostic to transmission technology, but should be able to enable further optimisations for cellular, optical and NTN communications, as well as for fog, edge and data centre computing environments. It offers a consistent/reliable programmable environment enabling “tailor made” implementation of various tenants’ requirements whilst providing secure and reliable scalability towards an unlimited number of requirements. The scope also covers new paradigms and solutions that are looking promising for the further reengineering of network architectures. Applicants should define the domain boundaries of their planned solutions and how they intend to maximise take up prospects in a fully heterogeneous domain. The focus of this Strand is on several complementary issues mentioned below and applicants may select one or more of those:

• AI powered edge cloud continuum. Support of a fully distributed and collaborative AI approach across the various RAN, edge and core CIC (Compute Inter-Connection) domains. An AI plane and associated functions (going beyond traditional Network Data Analytics) will optimise control and user plane operations in scenarios requiring diverse set of contradictory requirements (for example deterministic latencies and energy efficiency), including the optimisation of the optimal operational integration of heterogeneous (e.g. legacy) networks with the 6G Service Based Architecture. Also, in scope are specific AI/ML mechanisms suitable for: i) the transient nature of resources in the IoT domain (links, devices) e.g. constrains of compute power, energy, and time; ii) guaranteed convergence of meaningful outcomes in swarm-alike environments, i.e. facing the availability of many yet individually weak agents; iii) interfaces, data models, and orchestration strategies able to explore federated learning architectures and platforms close to the edge, to enhance data protection, improve inference reliability, and increase autonomy of end clusters. Energy efficient AI is in scope.
• Technologies for efficient Network and Service Resource Management in dynamic multi-tenant environments. This covers control and management aspects such as runtime service scheduling, conflict avoidance, conflict resolution, and the relationships between functions being executed in the deep edge (terminal or IoT device), the operator edge, and the core. These technologies include cross-domains solutions, on the fly SLA, architectures, as well as associated protocols, and algorithms, for dynamic, runtime assignment of resources to tasks, such that the executing system handles each task successfully under that task’s specific constraints while explicitly accounting for the resources used by the solution per se and its novel, added constraints. In scope are protocols and algorithms for user-to-system interface, i.e. exposing available resources and capabilities to the user applications and getting requirements from user applications explicitly or implicitly. It must achieve overall improvements in relevant KPIs (e.g. successful service throughput on an unchanged system; or resource usage such as energy, capacity, etc.) while avoiding (or resolving) potential conflicts brought by the potential uncoordinated usage of highly volatile resources, where the executing system strives to handle tasks successfully under that task’s specific constraints in such multi-tenant environments. Also in scope are the extreme challenges for IoT-oriented architectures, with challenging design given the particularities of its domain-specific resources (constrained battery-driven devices: security constraints, capability constraints, price limits), typical scaling and geographical spread expectations, and challenges for NTN environments.
• Energy efficiency enablers. In scope are protocols, algorithms, models, and policies to increase energy efficiency of the network. This covers metrics to capture energy consumption of resources in highly distributed, virtualised environments, including instrumentation to query and collect energy consumption metrics; models for target costing in terms of energy requirement per task; models able to specify the relationship of energy consumption with service and system KPIs and KVIs; policy definition and implementations for energy efficiency management in function of system requirements. Adherence to, and evolution over, ongoing standardisation efforts will be welcomed. The target is to provide the mechanisms to eventually realise the best energy aware dynamic implementation of network and service functions as a function of the use-case requirements.
• Pervasive Resilient Autonomic Resource Control in Virtualised Systems: In CIC networks, the multi-tenancy and dynamicity of the resource pool endangers the essence of the network existence – it is necessary to build a reliable and stable system with a (potentially very large) set of unreliable components. In scope are highly scalable, distributed, self-organising control and management protocols to provide in-band connectivity between all resources. These routing protocols should work across a variety of different topologies (sparse, dense, changing), should support mobility and multi-homing of nodes and avoid creating traffic concentrations. These protocols have to be used in zero-configuration and zero-touch approaches, essential so that no configuration errors can break the connectivity, and that failures be auto-corrected.
• Integrated and dependable sensing & actuation networks. The increasing penetration of the digital and physical world, in a new cyber reality, brings particular challenges for the reliability and trust of such systems, highlighting the need for new architecture concepts. The scope covers Integrating sensing and communication with the aim of making such functionalities available to users or operators of networks, e.g. how to expose any possible trade-offs, how to properly express access rights, etc., in view of addressing essential aspects for societal privacy and trust concerns, associating actions to frameworks promoting reliability and security. It presents extra complexity every time there is a complementary actuation when a network triggers actions in the real world. This includes issues such as exactly-once semantics, dependable execution of such actions, checking whether actions have indeed taken place by a corresponding sensing action and where issues like AAA and cost/billing for such activities needs to be addressed. The incorporation of these features in sematic oriented communications is also in scope.
• Digital network twinning applied in 6G: This includes the dynamic virtual representation of critical components and systems, including the simulation and modelling tools for large-scale real-time environments; derivation of network models (digital twins) from traffic analysis; and digital twin models as a core for network planning, management and control.
• New Communication Paradigms with enhanced intelligence. The work addresses innovative protocols in view of overcoming known Internet limitations as originating from new scenarios and vertical requirements (ultra-low latency, extreme mobility, ultra-high data rates, integration of end-terminals, controlled security, space applications). It considers systems where edge, access and cloud are increasingly undistinguishable (i.e. used homogeneously by the service layer), and diverse techniques can be considered, such as beyond IP networking or semantic/knowledge-based communications, in such fast-changing environment. Research should address functional improvements of the basic communication concepts, including transport mechanisms with improved packet delivery and/or energy efficiency able to cope with increased dynamics in network topologies. These communication paradigms should be able to flexibly operate in local/global architectures and provide primitives to perform the integration of new localized environments in an intelligent ICT system. Energy/efficiency improvements should be demonstrated in the research.