What is NWDAF?
The 5G Network Data Analytics Function (NWDAF) is a critical component in the 5G Standalone (SA) Core network, designed to deliver real-time insights that fuel network automation and enhance service orchestration. As an independent network function within the 5G Core architecture, NWDAF handles large-scale data analytics by collecting, processing, and analyzing data from multiple network sources. Defined by 3GPP standards, NWDAF plays a pivotal role in optimizing network performance through its interaction with key network components via the Nnwdaf (NWDAF Network) and Nnwdaf Management (NM) interfaces.
NWDAF’s primary functions include managing subscriptions, event exposure, and providing data analytics to consumers. By subscribing to event notifications and leveraging APIs, NWDAF gathers and analyzes data related to User Equipment (UE), Service Experience, and Network Load and Performance. These analytics serve a wide array of use cases, such as monitoring abnormal traffic patterns, evaluating network performance, and tracking UE behavior.
NWDAF Terminology
NWDAF is a key component in the 5G Standalone (SA) Core network, designed to harness network data analytics to provide real-time insights that drive network automation and optimize service orchestration in 5G networks.
NWDAF’s role is to manage and analyze large volumes of data collected from multiple network sources. As specified by 3GPP, NWDAF functions as an independent entity within the 5G Core architecture, interacting with other network components through established interfaces like the Nnwdaf (NWDAF Network) interface and the Nnwdaf Management (NM) interface.
The NM interface facilitates communication between NWDAF and the Nnwdaf Management Function (NMF), which oversees NWDAF instance management, analytics policy configurations, and performance monitoring.
Nwdaf Data Sources and Interactions
Network Functions that support the data analytics framework offers a service-based API for event exposure. NWDAF subscribes to these events, uniquely identified and published by Network Functions, enabling it to gather, analyze, and share data with relevant subscribers.
NWDAF interacts with multiple entities for the following purposes:
- Subscription-based data collection from AMF, SMF, PCF, UDM, and AF (directly or via NEF).
- Retrieval of data from repositories (e.g., UDR via UDM for subscriber-related information).
- Accessing information about network functions (e.g., NRF for NF-related details).
- On-demand and bulk analytics data provision to consumers.
- Data collection and analytics through the DCCF (Data Collection Coordination Function), which includes OAM data.
Figure 1: NWDAF Data Sources
The Nnf interface is key for NWDAF to manage data subscriptions, cancel subscriptions, and request specific data reports based on defined contexts.
NWDAF Analytics Categories
NWDAF categorizes its analytics into three primary groups:
- User Equipment (UE)
- Service Experience
- Network Load and Performance
These categories can be further refined by UE, groups of UEs, and slice applications.
NWDAF Use Cases
NWDAF Use Cases for Network-Related Analytics:
NWDAF supports two primary network performance events:
- “num_of_ue”: Tracks the number of attach requests within a specified time window.
- “sess_succ_rate”: Monitors session success rates over a designated time frame.
Consumers of NWDAF services can also subscribe to the “abnormal_behaviour” event with an exception ID like “unexpected_large_rate_flow” to receive updates on potential abnormal traffic patterns.
Figure 2: NWDAF Network-Related Analytics Use Case Flow
NWDAF Use Cases for UE-Related Analytics:
- UE Communications (“ue_comm”): Analyzes the volume of packets and bytes exchanged in uplink and downlink directions per PDU session.
- UE Mobility (“ue_mobility”): Tracks the cell ID associated with a specific UE.
Figure 3: Clause- Based Analytics Overview
NWDAF Deployment
Each NWDAF instance is identified by an analytics ID and its associated area of interest. A single NWDAF can handle multiple analytics IDs, with the area of interest representing the geographical scope of the network function.
In cases where multiple NWDAF instances are deployed within a Public Land Mobile Network (PLMN), these instances can function centrally, in a distributed fashion, or as a hybrid. Some NWDAFs may also act as aggregators, collecting analytics from other NWDAFs across various areas to produce consolidated reports.
NWDAF Data Analytics Functions
3GPP outlines several standard functions to support data analytics in 5G deployments:
- AnLF (Analytical Logical Function): Performs data inference, generating analytics (e.g., statistics or predictions) based on consumer requests, and exposes services such as Nnwdaf_AnalyticsSubscription or Nnwdaf_AnalyticsInfo.
- MTLF (Model Training Logical Function): Trains ML models and offers new training services.
- DCCF (Data Collection Coordination Function): Manages data collection and distribution to avoid redundant data requests.
- ADRF (Analytical Data Repository Function): Provides storage and retrieval services for data and analytics.
- MFAF (Messaging Framework Adaptor Function): Distributes analytics or event notifications across the network.
Figure 4: NWDAF Analytics ID
The architecture enables NWDAF and DCCF to collaborate on data management, while the MFAF and ADRF facilitate data transfer and storage.
Figure 5: 3GPP Release 16 NWDAF Analytics Provision Lifecycle
NWDAF and Machine Learning Model Deployment
NWDAF’s MTLF handles model training and deployment, while AnLF provides the historical data necessary for model predictions. DCCF coordinates data management, with MFAF and ADRF supporting data transfer and storage processes.
Figure 6: NWDAF ML Model Deployment
NWDAF Enhancements in Release 17
In 3GPP Release 17, new reference architectures were introduced to collect and expose UE data. These frameworks can be adapted for various data domains, providing flexibility for diverse analytics scenarios.
For UE data collection, the Application Function (AF) receives data reports from Data Collection Clients via PDU Sessions, exposing events to NWDAF through the Naf_EventExposure interface, as specified by SA2 and CT3 in TS 29.517.
Figure 7: NWDAF Analytics Event Provision
The scope of UE data collection and reporting is broadening, incorporating:
- Application Server (AS) instances as additional sources of UE data, where application logs can be analyzed for trends.
- Third-party ASPs, which can use the Data Collection AF for independent analytics aimed at enhancing their services.
taS
References:
- TS23.288 5G System; Network Data Analytics Services
- TS23.501 5G System Architecture
- TS29.517 5G System; Application Function Event Exposure
- TS26.531 Data Collection and Reporting; General Description
- TS26.532 Data Collection and Reporting; Protocols and Formats