AI RAN

Open6GE brings AI directly into the RAN scheduler, enabling per-UE intelligence, predictive link adaptation and dynamic scheduling based on real-time network conditions.

Building on the TOSSI OCUDU AI-RAN framework, the stack supports ML-based MCS selection, BSR prediction and CSI prediction, while extending intelligent scheduling towards different network slices with dynamically changing performance requirements and priorities. The approach combines offline training, lightweight in-RAN inference, live model updates and safe fallback to conventional scheduling.

Instead of relying entirely on static scheduling rules, future RAN systems can combine real-time network information with learned models while retaining safe fallback mechanisms. This creates a practical evolution path toward increasingly AI-native radio networks.