Network Data Analytics Function

In Release 15, the NWDAF (Network Data Analytics Function) was newly introduced as a defined architectural function within the 5G core network, as specified in TS 23.501. It was established to provide network data analytics services, such as network slice load predictions, which other network functions like an NSCE server can subscribe to via the NEF using procedures like Nnef_AnalyticsExposure_Subscribe. This enables capabilities like analytics exposure for monitoring and optimizing network slice performance based on collected data and predictions.

• Moving Network Analytics functionality into 23.501 TS 23.501CR0058
• Defining NWDAF in 23.501 TS 23.501CR0209
• Moving NWDAF to 23.501 TS 23.503CR0016
• Correct NWDAF resource TS 29.520CR0014

In Release 16, the NWDAF's role was significantly expanded to enable new, analytics-driven network automation. Key enhancements included using its analytics for specific functions like SMF and UPF selection, determining MICO mode parameters, and influencing UE mobility and background data transfer policies. Furthermore, the release formalized its exposure to external applications via the NEF and introduced new analytics services, such as network slice load predictions and performance monitoring, which other network functions could consume.

• Use of analytics for SMF selection TS 23.501CR0940
• Use of NWDAF analytics for decision of MICO mode parameters TS 23.501CR0837
• Use of analytics for user plane function selection TS 23.501CR0899
• Use of analytics for UE mobility procedures TS 23.501CR0900
• NEF service for NWDAF analytics TS 23.501CR0964
• CR for TS 23.501 Clarifications NWDAF Discovery and Selection TS 23.501CR0987

+ 41 more changes

In Release 17, NWDAF enhancements included new analytics applications such as using Redundant Transmission Experience analytics for URLLC services and extended UE Mobility analytics for LADN services. The release also expanded NWDAF's role in network slice management, enabling slice restriction and optimization based on its load predictions, and introduced procedures for NWDAF discovery and selection, including for model sharing and using provided ML models. Furthermore, it defined mechanisms for analytics exposure, allowing entities like the NSCE server to subscribe to NWDAF predictions via the NEF for network slice performance monitoring.

• Enchantments for supporting Supported Analytics Delay mechanism TS 23.501CR2530
• Network Slice restriction based on NWDAF analytics TS 23.501CR2567
• NWDAF discovery and selection TS 23.501CR2575
• NWDAF Discovery TS 23.501CR2577
• Adding the usage of Redundant Transmission Experience analytics for URLLC service TS 23.501CR2581
• Adding the usage of extended UE Mobility analytics for LADN service TS 23.501CR2582

+ 187 more changes

In Release 18, NWDAF enhancements focused on enabling analytics for roaming scenarios by introducing new NWDAF services and a reference point between NWDAFs, and on supporting federated learning through discovery and selection mechanisms for NWDAFs with that capability. The release also expanded NWDAF-assisted policy decisions, such as for URSPs (User Route Selection Policy), and extended its analytics exposure to functions like the PFDF (Policy and Charging Function Data Function) for application detection. Furthermore, NWDAF's registration information was extended to reflect new accuracy checking capabilities, and its data collection support was broadened for new analytics types like end-to-end data volume transfer time.

• Discovery and Selection of the NWDAF Supporting Federated Learning in 5GC TS 23.501CR3772
• NWDAF discovery principle enhancements for enhanced model sharing TS 23.501CR3783
• Discovery and selection of NWDAF with FL support - Resolve EN TS 23.501CR4070
• Using network analytics for roaming scenarios TS 23.501CR4080
• Update NEF to support NWDAF-assisted application detection TS 23.501CR4105
• Extension of NWDAF registration information to reflect new accuracy checking capability TS 23.501CR3764

+ 219 more changes

In Release 19, the NWDAF introduced new analytics capabilities including support for collision detection, location-related UE group analytics, and application layer AI/ML member capability analytics. It also expanded its scope with analytics for non-terrestrial access UE RAT connectivity, VAL performance for tethered UEs, and DN energy analytics. Furthermore, the release enhanced existing analytics for location accuracy, application performance, edge load, and slice usage patterns, while also defining new procedures for NWDAF model provision for AI positioning and updates to its discovery and selection parameters.

• Edge computing preparation analytics TS 23.436CR0030
• Support of Collision Detection Analytics TS 23.436CR0037
• Support of Location-related UE Group Analytics TS 23.436CR0038
• Support of Application Layer AI/ML Member capability Analytics TS 23.436CR0039
• UE RAT connectivity analytics for non terrestrial access TS 23.436CR0040
• Support for VAL performance analytics for tethered UEs TS 23.436CR0044

+ 89 more changes

In Release 20, key enhancements for NWDAF included new analytics for abnormal user plane traffic and traffic patterns, enabling policy control and mitigation actions based on these insights. The release also introduced support for monitoring the correctness of machine learning-enabled analytics. Furthermore, NWDAF's role was expanded to provide network slice load predictions to entities like the NSCE server, which can subscribe to these analytics via the Nnef_AnalyticsExposure_Subscribe service to trigger network slice optimization.

• Support monitoring ML-enabled analytics correctness TS 23.436CR0064
• Mitigation actions based on New Abnormal user plane traffic Analytics TS 23.501CR6507
• Policy control based on abnormal user plane traffic analytics and traffic pattern analytics TS 23.503CR1599
• Policy control based on new Abnormal User Plane Traffic Analytics TS 23.503CR1613
• Correction on QoS and policy assistance analytics TS 23.503CR1627