BubbleRAN, Telenor & NVIDIA Unveils Intelligent Network Slicing at MWC 2025

BubbleRAN, Telenor Research & Innovation demonstrating world’s first 5G Intelligent Network slicing empowered by NVIDIA’s AI Enterprise software to deliver private 5G maritime communication.

AI-powered network management

At MWC 2025 Barcelona, BubbleRAN is showcasing in collaboration with Telenor Research & Innovation and NVIDIA agentic AI-based Intelligent network slicing technology for private 5G. Network slicing is a technology that enables telecom operators to guarantee quality of service or specific applications or users, enabling them to offer premium services to enterprise customers. In current software implementations, a rules-based approach is used to create dedicated slices across access, transport and core. With advances in generative and agentic AI, complex reasoning and decision making is now possible for large-scale and dynamic data sets, making these technologies perfectly suited for a use case like network slicing. BubbleRAN’s intelligent network slicing solution leverages AI to significantly improve network experience and reduce operating expenses.

Telenor Research & Innovation, the experimental arm of the leading mobile network operator in the Nordics, is exploring differentiated services for private 5G networks in maritime transportation. Isolating network resources for different usage profiles is critical in applications where non-identical classes of service are involved.

The intelligent network slicing solution, developed by BubbleRAN leverages NVIDIA NIM microservices (for Both Llama 3.1 70B Instruct and Llama 3.1 8B Instruct) part of the NVIDIA AI Enterprise platform, and NeMo microservices to create and train large telecom models (LTMs). The LTMs provide foundation functionality for network agents that not only understand different traffic types but the optimizations required to support the desired quality of service across the end-to-end network. These agents actively observe the network states and metrics and continuously apply reconfiguration, management, and control actions to meet the desired quality of service. The LTM agents also offer a conversational interface to network engineers and are able to access various datasets through APIs to provide contextual answers related to network performance, service intent and suggested actions.

This groundbreaking technology greatly improves the quality of service for end users and streamlines joint RAN and core network slicing by means of a Kubernetes Slice Operator. This slice-aware engine replaces the pre-configured schedulers particularly in the radio access domain, enabling a more dynamic approach to network slicing. In this maritime pilot, three slices are autonomously deployed, operated, and their resources are dynamically allocated over time as their requirements evolve.”

Leadership Insights

In a cruise ship environment, we want to ensure that mission-critical traffic flows are treated with priority in the presence of other traffic types such as robotic remote inspection information, and broadband data for the passengers. This scenario requires an effective implementation of network slicing of the cruise private network.

– Senior Researcher Geir Egeland, (Telenor R&I) and CTIO Knut Fjellheim (TelenorMaritime)

Through this collaboration, network automation can reach a level never seen before in telecom.

– Navid Nikaein, CEO, BubbleRAN

Innovation

BubbleRAN’s Intelligent Telco Agent technology scales beyond network slicing to transform daily network operations with significantly improved accuracy, reduced computational overhead, and lower inference latency. “By integrating AI-driven intelligence into 5G, we are not just improving efficiency—we are transforming mobile network operations,” he added.