AI/ML in 3GPP 5G Advanced - Services and Architecture
Pradnya Taksande 1, Shwetha Kiran 2, Pranav Jha 3, Prasanna Chaporkar 4 · 2025
Read AI/ML in 3GPP 5G Advanced - Services and Architecture by Pradnya Taksande 1, Shwetha Kiran 2, Pranav Jha 3, Prasanna Chaporkar 4 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-03728.txt.
Introduction
Mobile networks are being rebuilt around artificial intelligence, and the global standards body 3GPP is writing the rules for that change. Pradnya Taksande and three colleagues survey the effort in a 2025 preprint on emerging communications technology. Their focus is Release 19 of the 5G Advanced standard, viewed through the work on service and system aspects.
The survey separates two directions of travel. In one, AI improves the network itself, for example by optimising how resources are used. In the other, the network is reshaped to serve AI applications such as image recognition. By organising the new features this way, the authors give engineers a map of what Release 19 adds. They also show where standardisation is heading next, as early work on Release 20 begins.
Catalog record
| CREATOR | Pradnya Taksande 1, Shwetha Kiran 2, Pranav Jha 3, Prasanna Chaporkar 4 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-03728
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
AI/ML in 3GPP 5G Advanced - Services and Architecture Pradnya Taksande 1, Shwetha Kiran 2, Pranav Jha 3, Prasanna Chaporkar 4 Abstract The 3rd Generation Partnership Project (3GPP), the standards body for mobile networks, is in the final phase of Release 19 standardization and is beginning Release 20. Artificial Intelligence/ Machine Learning (AI/ML) has brought about a paradigm shift in technology and it is being adopted across industries and verticals. 3GPP has been integrating AI/ML into the 5G advanced system since Release 18. This paper focuses on the AI/ML related technological advancements and features introduced in Release 19 within the Service and System Aspects (SA) Technical specifications group of 3GPP. The advancements relate to two paradigms: (i) enhancements that AI/ML brought to the 5G advanced system (AI for network), e.g. resource optimization, and (ii) enhancements that were made to the 5G system to support AI/ML applications (Network for AI), e.g. image recognition. I Introduction Artificial Intelligence (AI) and Machine Learning (ML) are transforming numerous industries and multiple aspects of modern life. From personalized recommendations on streaming platforms to real-time fraud detection in banking, AI/ML technologies are driving smarter decision-making across industries. In retail, they assist in inventory and supply chain management. In transportation, automotive vehicles rely on ML for object detection and navigation. Businesses leverage AI/ML for predictive analytics, customer insights, and operational efficiency. As data continues to grow, these technologies are evolving rapidly, reshaping how we work, interact, and solve complex problems, making them central to innovation in today’s world. AI/ML is playing a transformative role in mobile networks to enhance task efficiency, optimize network management and configuration, predict fault detection, and improve end-user experience. The 3rd Generation Partnership Project (3GPP), a global body responsible for mobile network standardization, began incorporating AI/ML features starting from Release 17 of the 5th Generation (5G) system. The 3GPP Standards development work is divided across three major groups: Radio Access Network (RAN), Service and System Aspects (SA), and Core Network & Terminals (CT). The SA group is responsible for standardizing Core Network (CN) functionalities, Security, Operations, Administration and Maintenance (OAM) functionality and end-to-end application support. The RAN group develops base station and radio interface standards, whereas the CT group develops User Equipment (UE) and CN protocols as well as interworking with external networks. Within 3GPP, AI/ML is being explored across multiple areas, including RAN intelligence, OAM intelligence, and incorporation of AI/ML capabilities in 5G Advanced architecture that began with Release 18. In 5G and beyond, plans are underway to support AI/ML related use cases like distributed computing for edge-cloud collaboration, energy-efficient operation at network as well as UE, autonomous network management, making mobile networks more adaptive, scalable, and resilient to changing demands. These use cases can broadly be divided into: 1) AI for Network : It includes the use of AI to optimize traditional algorithms, network functions, as well as network operations to improve performance, efficiency, and end-user experience. For instance, AI/ML can be used to design an efficient network resource management algorithm. 2) Network for AI : Here, the network needs to provide capabilities like communication with Quality of Service (QoS) guarantees, computing and storage resources to support AI/ML applications. For instance, when running an image recognition AI/ML application on the UE, the low latency connectivity to the edge or cloud needs to be provided by the network for additional processing support. Fig. 1 : Figure shows an autonomous car connected to 5G network. The car uses hierarchical AI/ML models. In-vehicle ML model manages real-time tasks like emergency braking and obstacle avoidance. For more complex and computation-intensive tasks, ML models are available at the edge and cloud nodes. Edge (cloud, resp.) models are used for moderate latency (non real-time, resp.) tasks like route optimization (optimization of battery management system, resp.). This hierarchical deployment ensures efficient task allocation based on latency and computational requirements. The ML model in the car may use data from its sensors, the model at the edge may use data from the cars in the same neighbourhood, and the model in the cloud can utilize global data. The data path in edge and that in cloud is shown with red and green dotted line, respectively. Number of surveys have been published summarizing potential of AI/ML for optimizing mobile networks. A comprehensive overview of AI and its role in communications within the context of 6G networks has been provided in [ 12 ] , however, the discussion on standardization aspects of AI/ML remains at a high level. [ 8 ] and [ 14 ] discuss AI/ML in network contexts, particularly related to RAN, but they do not focus on standardization. [ 10 ] primarily focuses on RAN standardization across various standardization bodies, viz. 3GPP and O-RAN. Similarly, [ 9 ] also emphasizes the RAN perspective but does not adequately cover the AI/ML activities carried out by 3GPP SA group. [ 11 ] provides a comprehensive overview of AI/ML technologies within the 3GPP RAN scope, including the proposed framework for applying AI/ML to the New Radio (NR) air interface. [ 13 ] provides an overview of Release 19 RAN enhancements in 5G-Advanced, purely from the 3GPP standardization perspective. At present, no literature specifically examines AI/ML capabilities introduced in 3GPP Release 19 from the SA group’s perspective. This article aims to bridge that gap. In the next section, we present challenges that need to be addressed while integrating AI/ML in networks. Subsequently, we provide a detailed account of AI/ML-related enhancements introduced in Release 19 within the 3GPP SA group. Finally, we present the AI/ML related study items approved for Release 20 to provide insight into AI/ML capabilities expected to be a part of 6G. II AI/ML in Mobile Networks AI/ML is playing an increasingly important role in mobile networks enabling smarter, efficient and adaptive networks, and also in various user applications. Thus, future networks are required not only to support AI/ML for its own efficient functioning, but also to provide communication, storage and compute support to AI/ML based applications. In this regard, 3GPP standardization looks at following two aspects: 1. AI-centric (Network for AI): This deals with the network efficiently supporting AI/ML functionalities like support for Federated Learning (FL), data and model exchanges between user applications and cloud servers, and providing desired QoS for communication requirements. See Fig. 1 for an illustrative example. 2. Network-centric (AI for network): This deals with efficient and secure acquisition, storage and management of data from various network elements, and then utilizing the data to train models that are integrated to improve networks’ efficiency. Next, we cover challenges in both scenarios mentioned above. 1. Hierarchical computing: Since mobile devices are usually battery-powered with limited computing and storage capabilities, processing complex tasks locally may not be feasible. To address this, the AI/ML workload can be split between the mobile device and an edge and/or cloud server. The challenge is in deciding how to split the ML model, which nodes to use, what are the Key Performance Indicators (KPIs) for each, how much computation power is required at each node, how much energy is consumed, and so on. 2. ML model distribution and sharing: The complexity of the ML model increases with the depth of the Deep Neural Network