Type something to search...
to navigateto selectESC to close
SMO Agent
SMO Agent

SMO Agent

Andrea Leone
Enrico Sbuttoni

The Service Management and Orchestration (SMO) Agent processes high-level intents for network lifecycle management, including deployment, monitoring, and deletion. It operates on top of the Hermes API Agent, translating service-level intents into structured orchestration requests.

develop
46
37

Detailed Description

The SMO Agent is part of the BubbleRAN MX-AI multi-agent architecture designed to automate network lifecycle management. It acts as the orchestration and intent-processing layer of the workflow. The agent translates high-level service intents into structured, executable requests for the Hermes API Agent, explicitly defining which APIs to invoke and which parameters to configure. While Hermes API executes validated Kubernetes operations, the SMO Agent focuses on intent interpretation, orchestration logic, and blueprint generation. This separation of responsibilities ensures modularity, operational safety, and clear abstraction between orchestration and execution layers.

Key Features

  • High-level intent processing and orchestration
  • Service blueprint generation for networks and terminals
  • Multi-agent workflow coordination
  • Hermes API orchestration layer

Use Cases

  • 5G network deployment orchestration
  • Network and UE lifecycle management

External Resources

AI Agent Training

BubbleRAN open documentation
View Details

Technical Details

Version2025.12
Published21 Jan, 2026
Base
Platform:
BubbleRAN MX-PDK and MX-AI

Authors

Andrea Leone
Andrea Leone
Enrico Sbuttoni
Enrico Sbuttoni

Tags

  • O-RAN
  • Artificial Intelligent
  • Monitor
  • Control
  • OpenAirInterface
  • srsRAN
  • Amarisoft
  • LITEON
  • Benetel
  • Open5GS

Affiliation

  • BubbleRAN
  • BubbleRAN
  • EURECOM

Certified By

  • BubbleRANBubbleRAN

More from Andrea Leone and Enrico Sbuttoni

AUTO-RAN Monitoring Agent

AUTO-RAN Monitoring Agent

Enrico Sbuttoni

An AI-driven monitoring agent designed to track key network KPIs, including UL/DL throughput, Signal-to-Noise Ratio (SNR), and pZeroNominal values for PUCCH and PUSCH, in cloud-native 5G O-RAN networks. The agent dynamically queries VictoriaMetrics and provides structured KPI responses to higher-layer agents within the multi-agent workflow.

O-RANMonitorArtificial Intelligent+1
develop
66
24
AUTO-RAN Validation Agent

AUTO-RAN Validation Agent

Andrea Leone

An AI-driven validation agent designed to evaluate power-related configuration parameters, including pZeroNominal for PUCCH and PUSCH, within an Amarisoft-based 5G network Digital Twin. The agent dynamically deploys a digital twin instance, executes controlled traffic simulations, and measures achieved uplink throughput to generate performance validation results.

O-RANMonitorArtificial Intelligent+2
develop
69
22
AUTO-RAN Config Planner Agent

AUTO-RAN Config Planner Agent

Enrico Sbuttoni

An AI-driven config planner agent that designs cell reconfiguration strategies to optimize network performance. It coordinates with the Monitoring Agent to retrieve KPI context and with the Validation Agent to assess candidate configurations before recommending an optimized reconfiguration plan to higher-layer agents.

O-RANArtificial IntelligentControl+1
develop
72
29
Hermes API Agent

Hermes API Agent

Enrico Sbuttoni

Hermes is an intelligent API agent for managing 5G O-RAN network resources through natural language interaction. It enables conversational deployment, retrieval, modification, and deletion of network entities such as terminals, networks, and infrastructure elements, with automatic YAML generation and schema validation for cloud-native network environments.

O-RANArtificial IntelligentMonitor+7
develop
18
10

Related Artifacts

SMO Agent Operational Intent Benchmark

SMO Agent Operational Intent Benchmark

Ilias Chatzistefanidis

A benchmark suite for LLM-powered multi-agent RAN assistants in a 5G Open RAN deployment. It evaluates 150 realistic operator intents (100 observability, 50 control) and reports coherence (0–5), action accuracy, latency, and GPU/VRAM usage across cloud and local LLM backends.

O-RANMonitorControl+2
release
84
66