Data Sovereign LLM-Assisted Automation Platform for Open Optical and Packet Transport Networks

The Open Optical Transport Network (OOPT) architecture, defined by the Telecom Infra Project (TIP), is currently being deployed by major operators. The OOPT ecosystem, with its multi-stakeholder and multi-vendor environment, faces challenges like interoperability, business-critical data sharing, and end-to-end control and automation. These issues are exacerbated by the introduction of AI-assisted network functions requiring real-time telemetry data. The integration of Large Language Model (LLM)-based AI agents further increases this complexity. This work proposes an experimentally validated automation platform that addresses these challenges. It leverages data spaces, carrier-grade data pipelines, and customized LLM assistants with NDA compliance. Our solution complies with OpenConfig and TAPI specifications for controlling programmable network elements and utilizes gNMI for telemetry streaming from open terminals and Optical Line System (OLS) controllers. We demonstrate the advantages of our proposal through several experiments on a large-scale, partially disaggregated packet-optical testbed. The experiments include autonomous transponder reconfiguration for optical link-capacity adjustment using AI-assisted traffic forecasting, LLM-assisted service provisioning, and policy-enforced telemetry sharing via data spaces.