Support of AI/ML for NG-RAN, as a RAN internal function, is used to facilitate Artificial Intelligence (AI) and Machine Learning (ML) techniques in NG-RAN.
The objective of AI/ML for NG-RAN is to improve network performance and user experience, through analysing the data collected and autonomously processed by the NG-RAN, which can yield further insights, e.g., for Network Energy Saving, Load Balancing, Mobility Optimization.
Support of AI/ML in NG-RAN requires inputs from neighbour NG-RAN nodes (e.g. predicted information, feedback information, measurements) and/or UEs (e.g. measurement results).
Signalling procedures used for the exchange of information to support AI/ML in NG-RAN, are use case and data type agnostic, which means that the intended usage of the data exchanged via these procedures (e.g., input, output, feedback) is not indicated.
AI/ML algorithms and models are out of 3GPP scope, and the details of model performance feedback are also out of 3GPP scope.
Support of AI/ML in NG-RAN does not apply to ng-eNB.
For the deployment of AI/ML in NG-RAN, the following scenarios may be supported:
-
AI/ML Model Training is located in the OAM and AI/ML Model Inference is located in the NG-RAN node;
-
AI/ML Model Training and AI/ML Model Inference are both located in the NG-RAN node.
The following information can be configured to be reported by an NG-RAN node:
-
Predicted resource status information;
-
UE performance feedback;
-
Measured UE trajectory;
-
Energy Cost (EC).
The collection and reporting are configured through the Data Collection Reporting Initiation procedure, while the actual reporting is performed through the Data Collection Reporting procedure.
The collection of measured UE trajectory and UE performance feedback is triggered at successful Handover execution.
Cell-based UE trajectory prediction, which can be used e.g. for the Mobility Optimization use case, is transferred to the target NG-RAN node via the Handover Preparation procedure to provide information for e.g. subsequent mobility decisions. Cell-based UE trajectory prediction is limited to the first-hop target NG-RAN node.