Components

Cognitive Compute Continuum

The Cognitive Compute Continuum allows for flexible and smart resource and service management by integrating Artificial Intelligence (AI). Specifically, an optimization algorithm enables the dynamic handling of computing, networking, and storage resources, participating in the management of a distributed environment. In 6G-BRICKS, an abstraction framework has been implemented to facilitate the flexible deployment, scaling, and migration of cloud native services, as network and resource conditions evolve. This framework is supported by two main components: the Edge Cloud Continuum Manager (ECCM), which oversees the orchestration of multiple edge clouds, and the π-Edge platform, which manages individual edge cloud domains. Together, they form the foundation of the Cognitive Compute Continuum, enabling more efficient and automated resource management in 6G networks. Specifically, ECCM’s primary role is to orchestrate the entire system efficiently by leveraging the intent driven optimization algorithm and Machine Learning (ML) predictions, while the π Edge platform is a cloud native edge management system designed to automate, maintain, and enable the interoperability of services throughout their lifecycle within Edge Cloud environments.

Key Features

01
Edge Cloud Continuum Manager (ECCM)
  • Infrastructure orchestration: management of the associated clusters
  • Application orchestration: optimization algorithm used for overall service and application Lifecycle Management (LCM)
  • 02
    π-Edge Platform
  • Cluster Manager: Interoperability with Kubernetes clusters and Docker Swarms and support for stateful service migration
  • MLOps: ML models for resource (CPU and memory) usage predictions of cluster’s nodes and event triggering
  • Management of X-Haul links: Dynamic activation/de-activation of a wireless link and exploitation of X-Haul switches as compute nodes
  • Technologies Involved

    The key technologies involved in the Cognitive Compute Continuum are related with the π-edge Platform. These are:

    • Docker Swarm and Kubernetes for the Cluster Manager.

    • Eclipse Mosquitto and RabbitMQ for the asynchronous communications between components and microservices of the platform.

    • Prometheus and Grafana for monitoring.

    • Kubeflow and Elyra for managing the ML workflows in MLOps component.

    • Graph Neural Network (GNN) models for resource usage predictions.

    • Wireless X-Haul system for connecting the Radio Access Network (RAN) and the Core Network (CN).

    Collaboration Process

    The Cognitive Compute Continuum components are deployed at the KUL testbed. Thus, a user can access its API and/or its GUI either directly (by hitting the respective URLs), assuming that they have access to the platform, or through the 6G-BRICKS portal.

    Scroll to Top