Page 6 - 3GPP_Highlights_I4

TECHNICAL HIGHLIGHTS

RAN3 COMPLETES

AI & ML SUPPORT STUDY

By Angelo Centonza,
3GPP Working Group RAN3 Vice-Chair

In September 2020, RAN plenary approved a new study to 3 main use cases investigated
investigate support for Artificial Intelligence and Machine
Learning (AI/ML) in 5G RAN architectures. It explored a In order to focus the effort on tangible solutions and to ensure
functional framework for RAN intelligence and identified convergence to well-defined guidelines for normative work,
use cases, based on the current 5G RAN architecture, where RAN3 identified three main use cases for which AI/ML based
the application of AI/ML techniques could bring substantial solutions would be investigated:
benefits. • Network Energy Saving: where the energy
consumption improvements for the whole radio access
The RAN3 study, concluded at plenary (RAN#95-e) in March network may be achieved by actions such as traffic
2022, was motivated by the strong interest shown by the 3GPP offloading, coverage modification and cell deactivation.
community in pushing RAN automation towards new targets, • Load Balancing: where the objective is to distribute
with the aim of achieving network optimisation decisions load effectively among cells or areas of cells in a multi-
in multi-variable scenarios for which classic, “rule-based” frequency/multi-RAT deployment to improve network
techniques would not be equally effective.
performance based on load predictions.
The approval of TR 37.817 - Study on enhancement for data • Mobility Optimization: where satisfactory network
collection for NR and ENDC marked a symbolic point in the performance during mobility events is preserved
history of 3GPP: standardisation moved from measurement- while optimal mobility targets are selected based on
fuelled processes to methods that make use of predictions. predictions of how UEs may be served.
Not only that, but such methods would also be able to predict
events within an identified event class, that were seldom - or
perhaps never - recorded before. One of the first steps taken by RAN3 was to identify key
principles on the basis of which technical solutions for each
This study was the first of its kind and its outcomes are use case could be developed. The following are some of the
influencing work in similar activities of other 3GPP WGs, such most relevant principles agreed:
as RAN1, SA5, and SA3.

AI/ML ALGORITHMS AND MODELS ARE IMPLEMENTATION >>> This principle ensures free model selection
SPECIFIC AND OUT OF STANDARDISATION SCOPE and the establishment of a framework that
fosters competition
Namely, the main scope of the study includes
SOLUTIONS DEVELOPMENT SHOULD FOCUS ON THE the definition of the corresponding types of
ENABLED AI/ML FUNCTIONALITY AND THE DEFINITION >>> inputs/outputs and on determining how to
OF THE CORRESPONDING TYPES OF INPUTS/OUTPUTS transfer needed information among
AI/ML functions, so to enable the desired AI/
ML functionality.
LOCATION OF AI/ML FUNCTIONALITIES WITHIN THE Namely, there would be flexibility on where AI/
CURRENT RAN ARCHITECTURE DEPENDS ON DEPLOYMENT >>> ML functionalities reside, depending on the
AND ON THE SPECIFIC USE CASES. use case and the solution selected.

MODEL TRAINING AND MODEL INFERENCE FUNCTIONS >>> Namely, information needed to run an AI/
SHOULD BE ABLE TO REQUEST DATA NEEDED FOR AI/ML ML based process may be provided on a
PURPOSES, IF NEEDED. subscription basis.

USER DATA PRIVACY AND ANONYMISATION SHOULD BE >>> It remains to be analysed whether specific
RESPECTED DURING AI/ML OPERATION. solutions are needed.

|
06 3GP P Highlights ne w slet t er