Proofs of Concepts

The 6G-INTENSE project will validate its key technologies and innovations through relevant Proof-of-Concepts (PoCs). Embracing technologies like AI/ML, edge computing, and intent-driven orchestration, the project employs cutting-edge platforms, such as the infrastructural platforms provided by Orange Romania, Eurecom and Athena/ISI. These platforms form the backdrop for a diverse set of experiments, exploring distributed continuum computing, metaverse services, and pervasive location awareness. With a commitment to fostering progress, 6G-INTENSE also places a significant focus on data acquisition, creating open datasets to fuel ongoing research in the dynamic 6G landscape. This holistic approach positions 6G-INTENSE at the vanguard of shaping the future of wireless networks. The PoCs are briefly presented below:

Distributed Continuum towards pervasive computing (PoC-1):

Objectives:

  • Manage smart services through the Network-Compute Fabric, trained by Generative AI models.
  • Ensure security in the Deep Edge infrastructure by monitoring links and performance.
Description:

Experiment 1.1 explores efficient media content distribution across diverse domains using a converged infrastructure. It employs dynamic Edge Cache instances for reliable streaming, adapting to user connectivity, content popularity, and network conditions. The 6G-INTENSE framework enables hierarchical, intent-driven automation for optimal Edge Cache services, including auto-scaling, auto-healing, and migration. INTRACOM’s content delivery network platform is deployed in various environments, and a multi-testbed setup showcases the INTENSE-6G Orchestration Continuum with πEdge local managers and orchestrators orchestrating Edge Cache container network functions.

Experiment 1.1 :

Pervasive Computing in a Distributed Continuum

Experiment 1.2 :

Edge intelligence and Compute interconnection

Objectives:
  • Efficiently connect network points using smart network technologies for enhanced video streaming.
  • Flexibly use resources within the system to optimize video streaming.
  • Ensure reliable quality of service through monitored intentions.
  • Improve video streaming quality by smartly training models on the edge.
Description:

Experiment 1.2 evaluates the Composable AI and learning capabilities of the 6G-INTENSE system in the context of distributed content delivery network video streaming. It introduces an automated service level objective setting and adjustment across domains, utilising Machine Learning as a Service (MLaaS) for domain adaptation. The experiment demonstrates automated resource brokerage, knowledge sharing, and cognitive enablement to ensure consistent service level agreement assurance across domains.

Metaverse (PoC-2):

Objectives:
  • Demonstrate the joint communication and sensing functions that provide pervasive location awareness to be leveraged by Metaverse.
  • Explore trade-offs on sensing accuracy vs. energy efficiency at the Deep Edge.
Description:

Experiment 2.1 focuses on intelligent algorithms for Joint Communication and Sensing in future 6G systems, specifically in user equipment tracking scenarios. While current algorithms excel outdoors, their precision diminishes indoors, impacting users connecting to Deep Edge devices via non-3GPP access. Sensing user location is crucial for predicting disruptions and enhancing location awareness. The scenario involves users connected through non-3GPP access to a 6G-INTENSE distributed intent-driven management and orchestration-deployed Metaverse service. Distributed intent-driven management and orchestration, relying on communication and sensing, showcases components translating users’ intents and demonstrating resource-level optimisation, such as handovers triggered by sensed information.

Experiment 2.1 :

Joint Communication and Sensing for optimal user tracking in the Metaverse

Experiment 2.2 :

Fully autonomous Metaverse fault, configuration, accounting, performance and security management, sensing, continuum abstraction

Objectives:
  • Showcase how Native AI mechanisms drive intent (re-) negotiation at the Tenant domain.
  • Deliver the Orchestration Continuum vision at the Service domain.
  • Demonstrate adaptation based on the inputs of a Sensing service that is part of the generalized Service Mesh.
Description:

Experiment 2.2 showcases distributed intent-driven management and orchestration end-to-end functionality, emphasising intent translation and propagation. Orange and Eurecom testbeds, interconnected via 6G-INTENSE domain management and orchestration, convey preferences through intent. A Metaverse Service Mesh aligns with 6G-INTENSE, featuring cloud-native network functions, micro-services, and a cyber-physical systems service. Domain management and orchestration autonomously handles onboarding and intent negotiation, while testbeds deploy services via their Network-Compute Fabric abstraction frameworks.

Local managers and orchestrators like Eurecom’s O-RAN and Intracom pi-Edge orchestrate resources. The experiment tests adaptability, intra- and inter-domain coordination, and conflict resolution, highlighting Native AI toolkit’s hierarchical reinforcement learning capabilities. Fault, configuration, accounting, performance and security management decisions include intent adaptations, service migrations, and resource scaling, illustrating collaborative orchestration by O-RAN and pi-Edge local managers and orchestrators.