Learn how 6G-Cloud validates the integration of the Cloud Continuum Framework, Management and Orchestration Framework and AI/ML Framework across the UMU and WUT testbeds.
From 6G-Cloud architecture to integrated Proof of Concept
This tutorial presents PoC-1, the first 6G-Cloud Proof of Concept, focused on the creation and validation of the project’s overall framework.
PoC-1 demonstrates how the main 6G-Cloud architectural building blocks can be deployed, interconnected and validated across heterogeneous testbed environments. It focuses on the interaction between three core frameworks:
- the Cloud Continuum Framework (CCF), responsible for resource abstraction, partitioning and orchestration across distributed cloud and edge environments;
- the Management and Orchestration Framework (MOF), responsible for the lifecycle management and orchestration of Network Services;
- the AI/ML Framework (AIMLF), responsible for AI-driven intelligence, model lifecycle management, inference and adaptation.
Together, these frameworks provide the foundation for validating how 6G-Cloud can support cloud-native, AI-driven and service-oriented 6G networks.
What you will learn
By the end of this tutorial, you will understand:
- how PoC-1 validates the integration of the main 6G-Cloud frameworks;
- how the CCF, MOF and AIMLF interact during runtime operation;
- how Resource Partitions are created and used to support Network Service deployment;
- how Network Services can be deployed and migrated across a cloud continuum environment;
- how AI Functions can be updated and integrated into orchestration workflows;
- how live data pipelines support model training, inference and closed-loop intelligence.
Why PoC-1 matters
The 6G-Cloud project proposes a service-oriented, cloud-native and AI-driven architecture for future 6G networks. To move from architecture design to practical validation, the project develops Proofs of Concept that demonstrate how its core components work in realistic environments.
PoC-1 plays a foundational role in this process. Rather than focusing on a single isolated function, it validates the interaction between the main architectural frameworks. This includes resource management, service orchestration, monitoring, AI/ML lifecycle support and cross-framework communication.
This makes PoC-1 an important step towards demonstrating that 6G-Cloud components can be deployed, interconnected and prepared for future KPI validation activities.
The PoC-1 architecture in brief
PoC-1 is implemented across two complementary testbeds.
At the University of Murcia, the Gaia Laboratory provides a heterogeneous cloud-edge environment with programmable compute clusters, virtualisation infrastructure, software-defined networking and multi-access connectivity. In this environment, PoC-1 focuses on runtime monitoring, data exchange, AI/ML workflows and cross-framework interaction between the MOF, CCF and AIMLF.
At Warsaw University of Technology, the PL-5G Lab Platform provides a multi-site 5G/6G experimentation environment. In this environment, PoC-1 focuses on Resource Partition creation, Network Service deployment, Network Service migration and AI Function update workflows.
Together, the two testbeds show how the 6G-Cloud architecture can be instantiated in different environments while preserving the same overall architectural principles.
Main components
Cloud Continuum Framework
CCFThe CCF provides the resource management layer of the architecture. It abstracts heterogeneous cloud, edge and far-edge resources and exposes them as a programmable resource pool. A central concept is the Resource Partition, a logically isolated subset of resources created to support a specific Network Service.
In PoC-1, the CCF is used to demonstrate resource discovery, Resource Partition creation, resource monitoring and orchestration actions such as workload migration.
Management and Orchestration Framework
MOFThe MOF manages the lifecycle of Network Services. It coordinates service deployment, monitors service and infrastructure state, and triggers orchestration actions when needed.
In PoC-1, the MOF interacts with the CCF to request resources and with the AIMLF to use AI/ML outputs that can support orchestration decisions.
AI/ML Framework
AIMLFThe AIMLF provides the intelligence layer of the architecture. It supports the lifecycle of AI Functions, including training, inference, performance monitoring and model updates.
In PoC-1, the AIMLF is demonstrated through workflows involving model training, inference, AI Function updates and integration with orchestration processes.
Video modules
Track A - WUT demonstration: Resource Partition and Network Service workflows
This module introduces the overall purpose of PoC-1 and explains how the WUT demonstration is structured. It provides the context for the following steps: initialisation, Resource Partition creation, Network Service deployment, Network Service migration and AI Function update.
This module shows the initialisation of the components required for the WUT demonstration. It prepares the environment for the Resource Partition and Network Service workflows that follow.
This module explains how a Resource Partition is created based on an intent received from the MOF. The CCF receives the request, identifies suitable resources, creates the required infrastructure and prepares the Resource Partition toolset.
This step is central to the 6G-Cloud cloud continuum concept, as it shows how resources can be selected and prepared to support a specific Network Service.
This module shows how the MOF deploys a Network Service on top of the Resource Partition. The Service Orchestrator retrieves the required deployment assets and applies them through the orchestration layer.
The module demonstrates how the 6G-Cloud architecture supports Network Service deployment across a multi-cluster environment.
This module demonstrates how the Network Service can be migrated according to a defined migration intent. The MOF Service Orchestrator generates the required orchestration policy, and the deployment is adapted across the available clusters.
This illustrates the ability of the system to support runtime adaptation, a key requirement for dynamic and distributed 6G environments.
This module shows how an AI Function can be updated during operation. In the WUT implementation, the AI/ML lifecycle uses MLflow and supporting deployment mechanisms to retrain, package and redeploy an updated AI Function.
This demonstrates how the AIMLF can contribute to adaptive intelligence within the 6G-Cloud architecture.
Track B - UMU demonstration: data pipelines and AI/ML workflow integration
This module focuses on the UMU implementation and shows how live data streams are created between the Network Service Director and the AIMLF. These data streams allow the AIMLF to collect operational metrics, train
This module presents the complete UMU demonstration. It focuses on the interaction between the MOF, CCF and AIMLF during steady-state Network Service operation.
The demo shows how operational data is collected, exposed and consumed by the AIMLF; how a model can be trained and used for inference; and how updated intelligence can be integrated into the orchestration loop without interrupting the running Network Service.
Step-by-step walkthrough
Frameworks are deployed in the testbed
The PoC-1 environment begins with the CCF, MOF and AIMLF deployed in the selected testbed. The CCF provides the resource management layer, the MOF provides the orchestration layer, and the AIMLF provides the AI/ML lifecycle and inference capabilities.
A Resource Partition is requested
The MOF requests the creation of a Resource Partition for a Network Service. The request includes the requirements that guide the selection and preparation of resources.
The CCF creates the Resource Partition
The CCF processes the request, identifies appropriate infrastructure resources and creates the Resource Partition. The partition is then prepared with the necessary tooling, monitoring and orchestration interfaces.
The Network Service is deployed
Once the Resource Partition is ready, the MOF deploys the Network Service. The deployment assets are retrieved and applied through the orchestration layer, allowing the service components to be instantiated across the selected resources.
Monitoring data is collected
During operation, the Network Service and the underlying infrastructure generate performance and resource metrics related to resource consumption, service performance and operational state.
Data is shared with the AIMLF
The AIMLF uses selected data streams to support model training, inference and online model replacement. This creates the basis for AI-driven orchestration and closed-loop decision-making.
AI Functions support orchestration
The AIMLF provides intelligence to the orchestration process by enabling predictions, model updates and AI Function lifecycle management.
Services and AI Functions are updated during runtime
The PoC-1 workflows demonstrate that both Network Services and AI Functions can be adapted during operation. Network Services can be migrated according to orchestration intent, while AI Functions can be updated through the AIMLF workflow.
Validation focus
PoC-1 supports the validation of the 6G-Cloud architecture by demonstrating that its core frameworks can be deployed, interconnected and operated together.
The tutorial focuses on functional validation, including:
- interaction between the CCF, MOF and AIMLF;
- Resource Partition creation and update;
- Network Service deployment and migration;
- exposure and use of monitoring data;
- AI/ML model training, inference and replacement;
- AI Function update workflows;
- preparation for future KPI evaluation.
The final quantitative KPI results will be reported in later project validation activities. PoC-1 establishes the technical foundation for these evaluations by confirming that the required workflows and framework interactions are operational.
Key takeaways
- PoC-1 shows that the 6G-Cloud architecture can be implemented across heterogeneous 5G/6G testbed environments.
- It demonstrates how the CCF, MOF and AIMLF can work together to support resource management, service orchestration and AI-driven intelligence.
- It also provides a practical foundation for validating future 6G capabilities such as cloud continuum resource sharing, proactive orchestration, energy-aware optimisation and AI/ML lifecycle automation.
Glossary
| Term | Meaning |
| CCF | Cloud Continuum Framework |
| MOF | Management and Orchestration Framework |
| AIMLF | AI/ML Framework |
| ResP | Resource Partition |
| NS | Network Service |
