AI-Native Cloud-RAN Orchestration for Enterprise Private 5G Using Digital Twin Models
Neha Tripathi, Krishna Kumar Gattupalli, Bhaksara Rallabandi, Monish Sai Medarametla
Abstract
Enterprise private 5G networks are appearing as an essential enabler to industrial automation, mission-critical communicational and smart campuses. However, their arrangement remains complicated because of the unequal devices, tough QoS requirements and working loads that are dynamic. Cloud-RAN (C-RAN) architecture provides scalability although the optimization in real time needs more pro- found intelligence. The framework of the enterprise Cloud- RAN proposed in this paper is AI-native, using Digital Twin (DT) models to model, forecast, and optimize the network performance in advance and deploy the live performance only after a simulation. Through intent-driven control through the integration of DT-driven forecasting and AI/ML-based policy orchestration, the framework can use closed-loop automation to avert zero-touch provisioning. The methodology is proved by the use-case analysis that covers predictive resource allocation, detecting anomalies and energy efficient scheduling. Our proposed framework and the conventional one minimize SLA violation up to 37 percent, increase throughput up to 22 percent, and save energy in RUs up to 18 percent. These results point to the distinction of AI-native orchestration that uses DT, meaning how enter- prise private 5G can be redesigned into a resilient, adaptable, and smart platform.