HARMONY
Project Profile

HARMONY

HARMONY brings agentic AI to O-RAN testing: a multi-agent system that autonomously manages, diagnoses, and reconfigures a live 5G SA testbed through natural language.

O-RAN Multi-Agent Systems 5G SA
View Repository

Agentic AI in Open RAN testing

HARMONY is a state-of-the-art multi-agent AI framework designed to orchestrate, diagnose, and self-heal 5G Standalone (SA) deployments under Open RAN (O-RAN) architectures. It acts as an autonomous operations controller, permitting network operators to communicate complex test sequences, slice definitions, and fault diagnostics through simple natural language.

How It Works

  • Specialized AI Agents: HARMONY splits core tasks across three distinct specialized agent types:
    1. Diagnostics Agent: Analyzes live metrics and log files from the central unit (CU) and distributed unit (DU) to find protocol faults.
    2. Configuration Agent: Updates radio configurations via standardized REST APIs and Netconf/YANG schemas.
    3. Verification Agent: Deploys virtual user equipment (UEs) to test throughput and packet latency, verifying configuration updates.
  • Autonomous Self-Healing Loop: If a fault is diagnosed, the agents cooperatively plan a rollback or mitigation sequence without manual engineering intervention.
← Back to all projects