Experimental Demonstration of Autonomous Class Aware Optimization via Digital Twin Aided Optical Network Agent
We demonstrate on a network testbed a Digital Twin-aided multi-agent LLM for autonomous class-aware optical network management. The system detects soft failures and autonomously adapts modulation format per service class requirements, ensuring continuous, marginless channel quality with minimal human intervention. ©2026 The
physics.optics↗