Work on AI is now well established in 3GPP, with studies and specifications expected in Release 18 that will move standards towards providing processes that make full use of the predictive power of data.
Both TSG RAN and TSG SA groups have drawn up specific descriptions of the features needed to embrace both Artificial Intelligence and its class-mate Machine Learning.
Working Group (WG) RAN3 has completed TR 37.817 – a Study on enhancements for data collection for NR and ENDC, focusing on three main use cases for AI/ML solutions:
- Network Energy Saving (traffic offloading, coverage modification and cell deactivation).
- Load Balancing, to distribute load effectively among cells or areas of cells in a multi-frequency/multi-RAT deployment to improve network performance based on load predictions.
- Mobility Optimization: where satisfactory network performance during mobility events is preserved while optimal mobility targets are selected based on predictions of how UEs may be served.
WG RAN1 has a Study (RP-213599) on Artificial Intelligence, Machine Learning for the NR Air Interface, looking at performance, complexity and potential specification impacts.
The RAN 1 use cases will focus on:
- Channel state information (CSI) feedback enhancement, e.g., overhead reduction, improved accuracy, prediction.
- Beam management, e.g., beam prediction in time,and/or spatial domainfor overhead and latency reduction, beam selection accuracy improvement.
- Positioning accuracy enhancements for different scenarios (e.g., those withheavy NLOS conditions).
This Study, and its resulting Technical Report (TR) in the 38 series, will lay the foundation for future radio air interface use cases leveraging AI/ML techniques.
At the system level – WG SA2 is preparing its Study, with a focus on 5GS architectural and functional extensions, to allow providers to offer the 5G System as a platform to support AI/ML-based services over the application layer. The SA2 study (TR 23.700-80), follows on from WG SA1’s identified requirements (in TS 22.261) for the support of AI/ML model distribution, transfer, training for various applications (e.g. video/speech recognition, robot control, automotive) for three main types of operations:
- AI/ML operation splitting between AI/ML endpoints.
- AI/ML model/data distribution and sharing over 5G systems.
- Distributed/Federated Learning (FL) over 5G systems.
Media and Management
Both WG SA4 - Multimedia Codecs, Systems and Services and WG SA5 - Management, Orchestration and Charging also have important studies underway:
- Study on Artificial Intelligence and Machine Learning for Media (SP-220328), to identify relevant interoperability requirements and implementation constraints of AI/ML in 5G media services.
- Study on AI/ML management (SP-211443), to study the management capabilities and management services to support and coordinate AI/ML in 5GS.
ENDC: E-UTRAN New Radio – Dual Connectivity
3GPP Work Plan:
See a listing of all work & study items here: https://www.3gpp.org/ftp/Information/WORK_PLAN/ and search for ‘AI’ in the acronym field of the Excel sheet there.
RAN1, RAN3, SA1, SA2, SA4, SA5
|950008||Study on AI/ML Model Transfer Phase2||FS_AIML_Ph2||Rel-19||SA1|
|940071||Study on 5G System Support for AI/ML-based Services||FS_AIMLsys||Rel-18||SA2|
|940084||Study on Artificial Intelligence (AI)/Machine Learning (ML) for NR air interface||FS_NR_AIML_air||Rel-18||RAN1|
|941010||Artificial Intelligence (AI)/Machine Learning (ML) for NG-RAN||NR_AIML_NGRAN||Rel-18||RAN3|
|940039||Study on AI/ML management||FS_AIML_MGMT||Rel-18||SA5|
IMPORTANT NOTE: Please be aware that these pages are a snapshot of the work going on in 3GPP. The full picture of all work is contained in the Work Plan (https://www.3gpp.org/ftp/Information/WORK_PLAN/)