Taiwan-based LITEON Technology, the Singapore University of Technology and Design (SUTD), and NeuroRAN have demonstrated new AI-RAN (Radio Access Network) technology for 5G networks.

The solution is targeted at smart city applications, including intelligent urban surveillance and industrial automation.

By integrating a distributed User Plane Function (dUPF) with the 5G core, the system reduces end-to-end latency for critical urban infrastructure monitoring.

The technology natively integrates AI into 5G radio infrastructure, allowing for real-time edge AI applications. The system uses dynamic AI model partitioning, which distributes workloads between user devices and the network edge based on real-time spectrum sensing and network conditions.

The companies claim this approach reduces latency and energy consumption while improving data privacy by keeping sensitive information on-device. The demonstration utilised NVIDIA AI Aerial software and high-performance computing to support real-time video object detection.

“We are accelerating AI-RAN from vision to commercialisation,” said Richard Chiang, General Manager of LITEON’s Smart Life Application Strategic Business Unit. “We aim to deliver more flexible and efficient network architectures for telecom operators and enterprises.”