OpenOP-KG: Knowledge Graph Driven Experiment Intelligence and Reproducibility Framework for ETSI OpenOP 6G Testbeds

The ETSI OpenOP initiative represents a significant step toward open, federated, and multi-domain 6G experimen tation. However, as these testbeds grow in complexity and scale, researchers face significant hurdles in experiment discovery, resource selection, and—most critically—reproducibility. Existing orchestration frameworks focus primarily on low-level resource management and deployment automation, leaving a ”semantic gap” in how experimentation knowledge is captured and reused. In this paper, we propose OpenOP-KG, a novel framework that introduces a semantic intelligence layer top of ETSI OpenOP. By modeling testbed resources, network capabilities, experimen tation workflows, and historical Key Performance Indicators (KPIs) within a unified Knowledge Graph (KG), OpenOP-KG enables intelligent experiment recommendation and automated reproducibility support. We define a semantic similarity model that allows researchers to discover reusable experiment templates and compatible infrastructure across federated domains. Further more, we introduce an AI-assisted ”Experiment Digital Memory” that captures systemic dependencies to ensure experiments re main reproducible even as underlying infrastructures evolve. Our evaluation, conducted using representative 6G experimentation scenarios, demonstrates that OpenOP-KG reduces experiment preparation effort by up to 45% and significantly enhances the reliability of cross-site experiment replication.