Gorde:
Xehetasun bibliografikoak
Egile Nagusiak: Karamchandani Batra, Amit, González-Sánchez, Daniel, de la Cal García, Luis, Bellido Triana, Luis, Mozo Velasco, Bonifacio Alberto, Lentisco, Carlos M., De la Osa Mostazo, David, Pastor Perales, Antonio, R. Lopez, Diego
Formatua: Recurso digital
Hizkuntza:
Argitaratua: Zenodo 2026
Gaiak:
Sarrera elektronikoa:https://doi.org/10.5281/zenodo.18717877
Etiketak: Etiketa erantsi
Etiketarik gabe, Izan zaitez lehena erregistro honi etiketa jartzen!
Aurkibidea:
  • This repository provides the ML inference engine for the ACROSS TC3.2 use case (Smart Energy-aware Zero-touch Traffic Engineering). The software predicts router power consumption from instantaneous network telemetry-derived metrics to support Network Digital Twin (NDT) environments and energy-aware traffic engineering. The inference service is containerized (Docker/Docker Compose) and uses Kafka for message exchange: it consumes input telemetry features, runs a selected ML model, and publishes predicted power consumption along with feature-based variation rates. The repository also includes a telemetry mock service to replay experiment datasets, validate predictions against ground truth, and generate verification plots. Supported router types: RA and RB. Supported model types: linear regression, polynomial regression, random forest, and deep neural networks (with optional scaling). Models are organized by router type/model type and selected dynamically via environment configuration.