Deployment of Multi-Agent System Pipelines for Near-Real-Time Operation of 6G Network Services

A Machine Learning Function Orchestrator deploying and reconfiguring Multi-Agent Systems (MAS) pipelines to support near-real-time control of network services is showcased. An architecture to orchestrate the dynamic deployment of agents has been defined by integrating different components. Near-real-time operation of an optical system controlled by a distributed MAS fed with telemetry is presented as use case.