Description
In 3GPP specifications, the term 'AI' serves as a standardized prefix for Application Interface class methods. It is a naming convention used within interface definitions to clearly identify methods that belong to the application layer of network functions. This prefix is applied across various technical specifications (TS) to maintain consistency in how application programming interfaces (APIs) and service-based interfaces (SBIs) are documented and implemented.
The AI prefix is typically found in the context of Open Service Architecture (OSA) and later in the Service-Based Architecture (SBA) of the 5G Core network. It precedes the actual method name, forming a complete identifier like 'AI_<MethodName>'. This structured naming helps in distinguishing application-layer operations from transport, session, or management-layer functions within complex network element interfaces. The methods themselves define the operations that an application can invoke on a network function or that network functions can expose to each other, such as service registration, discovery, invocation, and policy management.
Architecturally, interfaces using the AI prefix are part of the broader framework for enabling third-party application interaction with network capabilities, as defined in specifications like TS 23.090 (Open Service Access) and TS 23.271 (Location Services). In modern 5G systems, this concept evolves into the Network Exposure Function (NEF) and standardized APIs, but the AI prefix historically provided a clear marker for application-relevant methods within interface specifications. Its usage ensures that during system design, code generation, and testing, engineers can easily identify and handle methods intended for application-level logic and service exposure.
Purpose & Motivation
The AI prefix was introduced to solve the problem of inconsistent and ambiguous naming for application-layer methods within 3GPP interface specifications. Prior to its standardization, different working groups and releases might use varied naming conventions (like 'App', 'Srv', or no prefix at all) for similar interface methods, leading to confusion during implementation and integration between network equipment from different vendors. The prefix creates a uniform, easily recognizable pattern that denotes a method's belonging to the application interface domain.
Its creation was motivated by the need for clear, maintainable, and interoperable specifications as 3GPP networks began to expose more capabilities to external applications, starting with initiatives like the Open Service Access (OSA) and Parlay/OSA APIs. By tagging these methods with 'AI', the standards body ensured that anyone reading a technical specification could immediately understand the functional layer of the operation, streamlining the development of network elements and client applications that rely on these standardized interfaces.
Classification
Detected Changes Across Releases
from 3GPP Change RequestsSpecific changes extracted from the „Change history“ tables of 3GPP specifications (36 CRs across 4 releases). Complements the general historical overview above with the evidence-based evolution of this function.
In Release 15, the standardization of the Application Interface was introduced as a defined interface for applications and clients to access network service capability features. This provides a framework for developing applications using standardized Application Programming Interfaces (APIs). The release also included a correction to the source transport network layer address parameter on the NG-C interface.
- Correction on source TNL ADDRESS in NG-C interface TS 38.423CR0170
In Release 17, the enhancements to the Application Interface (AI) primarily involved terminological corrections for AI/ML-related terms and provided clarifications on the application of slice-based Random Access Channel (RACH) configuration procedures. These updates served to refine the standardized interface used by applications to access network service capability features. The changes focused on improving the precision of technical documentation without introducing new architectural capabilities.
In Release 18, the enhancements for the AI/ML management function introduced new standardized interfaces and procedures specifically for AI/ML in the NG-RAN, including support for split architecture deployments. The updates focused on architectural enhancements to facilitate communications, refining the Application Interface for accessing these new AI/ML service capability features. This included corrections and definitions for AI/ML terminology and management within the NRM framework.
- Add relations for NRMs related to AI/ML inference capabilities TS 28.104CR0079
- Enhancements for AI-ML management TS 28.105CR0076
- AI/ML for NG-RAN TS 38.300CR0756
- Support of AI/ML in NG-RAN in the case of split architecture TS 38.401CR0265
- Support of AI/ML in NG-RAN TS 38.423CR0959
- CR Rel-18 TS28.105 AI/ML management TS 28.105CR0151
+ 13 more changes
In Release 19, the AI/ML function saw enhancements focused on positioning accuracy and RAN integration. Specifically, new support was introduced for AI/ML-assisted positioning and for AI/ML features over the NR Air interface. Furthermore, the release provided specific enhancements and corrections for AI/ML within the NG-RAN architecture and for AI/ML-based Coverage and Capacity Optimization (CCO).
- Introduction of AI/ML Positioning Accuracy Enhancements TS 37.355CR0559
- Introduction of AI/ML for NR Air interface feature in TS38.300 TS 38.300CR1006
- Support of enhancements on AI/ML for NG-RAN TS 38.300CR1033
- Support of AI/ML assisted positioning TS 38.401CR0477
- Enhancement of AI/ML for NGRAN split architecture TS 38.401CR0478
- Support of enhancements on AI/ML for NG-RAN TS 38.423CR1411
+ 8 more changes
Explore further
Broader topics and technologies where AI plays a role.
Defining Specifications
3GPP specifications that define or reference AI, with the latest known release. Sourced from the 3GPP document catalog — see methodology.
| Specification | Title | Release |
|---|---|---|
| TR 21.905 vj00 | 3GPP Technical Terms and Definitions | Rel-19 |
| TS 22.156 vj10 | Mobile Metaverse Services | Rel-19 |
| TR 22.829 vh10 | Enhancement for UAVs; Stage 1 | Rel-17 |
| TR 22.856 vj20 | Feasibility Study on Localized Mobile Metaverse Services | Rel-19 |
| TR 22.873 vi00 | Technical Report on IMS Multimedia Telephony Service Enhancements | Rel-18 |
| TR 22.874 vi20 | Technical Report | Rel-18 |
| TR 22.890 vj00 | Study on Railway Smart Station Services | Rel-19 |
| TS 23.090 vj00 | USSD Stage 2 Specification | Rel-19 |
| TS 23.171 v1300 | LCS Stage 2 Specification for UMTS | Rel-4 |
| TS 23.271 vj00 | LCS Stage 2 Specification | Rel-19 |
| TS 23.700 vk00 | XR Services Application Enablement Layer | Rel-20 |
| TS 25.211 vj00 | UTRA FDD Layer 1: Transport & Physical Channels | Rel-19 |
| TS 26.847 vj00 | AI/ML Evaluation in 5G Media Services | Rel-19 |
| TS 26.854 vj00 | Study on Haptics in 5G Media Services | Rel-19 |
| TR 26.927 vj00 | AI/ML in 5G Media Services Study | Rel-19 |
| TR 26.928 vj00 | Study on eXtended Reality (XR) in 5G | Rel-19 |
| TR 26.956 vj01 | Beyond 2D Video Formats & Codecs Study | Rel-19 |
| TS 28.104 vj30 | Management Data Analytics (MDA) | Rel-19 |
| TS 28.105 vj30 | AI/ML Management for 5GS | Rel-19 |
| TR 28.809 vh00 | Enhancement of Management Data Analytics (MDA) Study | Rel-17 |
| TS 33.784 vj00 | Security aspects of AI/ML in core network | Rel-19 |
| TR 33.877 vi00 | Technical Report on Security Aspects of AI/ML in RAN | Rel-18 |
| TR 33.898 vi01 | Technical Report on 5GS AI/ML Security | Rel-18 |
| TS 37.340 vj00 | Multi-Connectivity Operation Overview | Rel-19 |
| TS 37.355 vj20 | LTE Positioning Protocol (LPP) | Rel-19 |
| TS 38.300 vj00 | NG-RAN Overall Description | Rel-19 |
| TS 38.305 vj00 | NG-RAN UE Positioning Stage 2 | Rel-19 |
| TS 38.401 vj10 | NG-RAN Architecture Specification | Rel-19 |
| TS 38.423 vj10 | Xn Application Protocol (XnAP) specification | Rel-19 |
| TS 38.843 vj00 | Study on AI/ML for NR Air Interface | Rel-19 |