Description
The Network Data Analytics Function (NWDAF) is a standardized analytics and machine learning-enabled function within the 5G Core network architecture, introduced in 3GPP Release 15. It serves as a centralized or distributed analytics engine that collects, correlates, and analyzes data from various network functions, applications, and external sources to generate insights, predictions, and recommendations. The NWDAF's primary role is to provide data-driven intelligence to other 5G Core control plane and management plane functions, enabling automated network optimization, proactive service assurance, and enhanced user experience. Architecturally, it is defined as a service-based architecture (SBA) function that exposes northbound analytics services via standardized APIs, allowing consumers like the Policy Control Function (PCF), Network Slice Selection Function (NSSF), and Operations, Administration and Management (OAM) systems to request analytics.
NWDAF operates through a structured data pipeline involving data collection, analytics processing, and result dissemination. It collects data from multiple sources categorized as network data (e.g., from AMF, SMF, UPF regarding session metrics, mobility patterns, QoS), application data (e.g., from Application Functions (AF) about service requirements), and management data (e.g., from OAM about performance and faults). This data can be ingested via push or pull mechanisms using standardized interfaces like Nnwdaf_EventsSubscription. Internally, the NWDAF employs analytics models—which can be rule-based, statistical, or machine learning models—to process the data. These models perform tasks such as anomaly detection, trend analysis, prediction of network load or user mobility, and identification of slice performance issues. The analytics results, which include historical reports, real-time insights, or future predictions, are then provided to consumer functions to influence decisions, such as adjusting QoS policies, triggering network slice reconfiguration, or optimizing handover parameters.
Key components of the NWDAF include the Analytics Logical Function (which hosts the analytics models), the Data Collection Coordination Function (which manages data acquisition from sources), and the Analytics Exposure Function (which handles API exposure and subscription management). The NWDAF supports both on-demand analytics queries and subscription-based continuous analytics reporting. For example, a PCF may subscribe to NWDAF for analytics on expected network congestion in a certain area to pre-emptively adjust policy rules, or an OAM system may request predictions on slice resource utilization for capacity planning. The NWDAF also facilitates model training and lifecycle management, potentially leveraging federated learning across multiple NWDAF instances. Its integration with the 5G Core is pervasive, impacting nearly all aspects of network operation from mobility management and session management to network slicing and edge computing, making it a cornerstone of 5G's autonomous network vision.
Purpose & Motivation
The NWDAF was created to address the growing complexity and dynamic nature of 5G networks, which support diverse services with vastly different requirements—from enhanced mobile broadband to massive IoT and ultra-reliable low-latency communications. Traditional network management relied on reactive, manual interventions based on static thresholds, which were inadequate for 5G's scale, agility, and service diversity. The NWDAF introduces a native, standardized analytics capability within the 5G Core to enable data-driven, proactive, and automated network operations. It solves the problem of siloed data across network functions by providing a centralized analytics hub that correlates information to derive holistic insights, thereby improving network efficiency, resource utilization, and service quality.
Historically, network analytics were performed by external OAM systems or proprietary solutions that lacked tight integration with the core network control plane. This resulted in delayed responses, limited real-time capabilities, and inability to directly influence network behavior. With the advent of network slicing, edge computing, and service-based architecture in 5G, the need for intelligent, closed-loop automation became critical. Release 15 of 3GPP laid the foundation for NWDAF as part of the 5G system's native intelligence, motivated by operators' demands for reduced operational costs, improved customer experience, and support for new vertical services. The NWDAF enables predictive maintenance, dynamic resource allocation, and personalized service delivery by leveraging machine learning and big data techniques within the telecom domain.
Furthermore, NWDAF addresses limitations of previous approaches by providing a standardized framework for analytics that ensures interoperability across vendors and network functions. It allows network operators to deploy advanced analytics without relying on fragmented, vendor-specific solutions. By embedding analytics directly into the network architecture, NWDAF facilitates real-time decision-making, such as instantly adapting to traffic spikes or mitigating congestion before it impacts users. Its creation was also driven by the vision of self-organizing networks (SON) evolving towards fully autonomous networks, where NWDAF acts as the 'brain' that analyzes data and triggers actions through policy control and management systems. This transforms 5G from a connectivity platform into an intelligent, adaptive infrastructure capable of meeting the stringent and varied demands of future digital society.
What changed in its specifications
The latest release-to-release comparison of specifications that define NWDAF. Each summary describes the whole specification, not only this term.
Rel-19 adds new clauses and APIs for analytics procedures. These cover data storage, edge computing preparation, collision detection, and location-related UE group analytics.
Full comparison in TS 23.436 →Rel-19 added MPQUIC steering to multi-access PDU sessions, XR and media handling, and an Energy Information Function. It also added Local Offloading Management for edge computing, Vertical Federated Learning for AI/ML, and Ambient IoT text.
Full comparison in TS 23.501 →Architecture
Where it sits in the network
- Mobile Network → 5G Core → NWDAF (Network Resource Management)
Technology family
Network exposure and analytics · Core Network
- Rel-11 SCS/AS
- Rel-13 IWK-SCEF, SCEF
- Rel-15 CAPIF, NEF, NWDAF
- Rel-16 SCEF+NEF
- Rel-17 ADRF, AnLF, DCCF, L-NEF, MDAF, MFAF, MTLF
- Rel-18 RE-NWDAF
- NWDAF decomposed into MTLF
“When one or more trained ML Models are available for one or more Analytics ID(s) the NWDAF containing MTLF shall include the Analytics ID(s) that is(are) supported per service in the registration towards NRF.” TS 23.288
- NWDAF decomposed into AnLF
“If the NWDAF service consumer needs to discover an NWDAF containing AnLF which can use the Model provided by the specific NWDAF containing MTLF” TS 23.288
- NWDAF extended by RE-NWDAF
“The NWDAF with roaming exchange capability is called Roaming Exchange NWDAF (RE-NWDAF).” TS 23.288
- NWDAF used together with DCCF
“The functionalities provided by DCCF and/or ADRF can also be hosted by an NWDAF.” TS 23.501
- NWDAF used together with ADRF
“The functionalities provided by DCCF and/or ADRF can also be hosted by an NWDAF.” TS 23.501
- NWDAF used together with MDAF
“Management Data Analytics Function (MDAF) may act as 3GPP domain-specific (e.g. RAN or CN) or as 3GPP cross-domain MDA MnS producer. Figure 5.2-1 illustrates the example of coordination between NWDAF, gNB and MDA MnS producer(s) for data analytics purpose.” TS 28.104
Members and relations were proposed independently by several models and kept only where at least two agree; quotes are verified verbatim against the specifications. Release = first approved version of the defining specification.
Release Timeline
Detected Changes Across Releases
verified against 3GPP sourcesDocumented changes in 6 releases, each verified against the text of 3GPP specifications or approved change requests and shown with the exact wording. The release timeline above shows the same releases. A release without an entry has no documented change for this term.
TS 29.520 appears for the first time, specifying the NWDAF Nnwdaf_EventsSubscription service procedures, and 23.501 now defines the NWDAF.
- “Upon the reception of an HTTP POST request with: "{apiRoot}/nnwdaf-eventssubscription/v1/subscriptions" as Resource URI” TS 29.520 v15.11.0
- Defining NWDAF in 23.501 TS 23.501CR0209
- “The NWDAF shall invoke the Nnwdaf_EventsSubscription_Notify service operation to notify the subscribed event.” TS 29.520 v15.11.0
TS 23.288 first specifies the NWDAF in Rel-16: deployment options, analytics provision to 5GC NFs and OAM, data collection, and subscription/request handling, plus the Nnwdaf event subscription API in TS 29.520.
- “A single instance or multiple instances of NWDAF may be deployed in a PLMN.” TS 23.288 v16.12.0
- “The NWDAF provides analytics to 5GC NFs and OAM as defined in clause 7.” TS 23.288 v16.12.0
- “When a subscription to analytics information is received, the NWDAF determines whether triggering new data collection is needed.” TS 23.288 v16.12.0
Rel-17 extended NWDAF with multiple-instance deployment, Aggregator NWDAF, MTLF/AnLF decomposition, UDM registration, and analytics subscription transfer with a Termination Request.
- “an NWDAF can act as an aggregate point (i.e. Aggregator NWDAF) and collect analytics information from other NWDAFs, which may have different Serving Areas” TS 23.288 v17.12.0
- “NWDAF can contain an MTLF or an AnLF or both logical functions.” TS 23.288 v17.12.0
- NWDAF registering into UDM TS 23.288CR0259
Rel-18 extends NWDAF with accuracy checking capability, federated learning server discovery and ML model provision, new analytics output and subscription handling, plus data collection from UPF, GMLC and OAM.
- “A NWDAF may have the accuracy checking capability for Analytics and/or ML Models.” TS 23.288 v18.13.0
- “the NWDAF containing MTLF should discover an FL server NWDAF as described in clause 5.2” TS 23.288 v18.13.0
- Enhance NWDAF to enable Federated Learning TS 23.288CR0634
NWDAF gained new Rel-19 capabilities: ML model training and provisioning for LMF-based AI/ML positioning, collecting input data from the LMF, and acting as a Vertical Federated Learning server or client.
- “To train an ML model or to do ML model performance monitoring for LMF-based AI/ML positioning, the NWDAF can collect input data from LMF” TS 23.288 v19.8.0
- “This clause specifies procedures for Vertical Federated learning where AFs and NWDAF can can either act as VFL servers or VFL clients.” TS 23.288 v19.8.0
- Support AI/ML model performance monitoring by NWDAF TS 23.288CR1253
NWDAF analytics accuracy handling was extended to include loss values for continuous outputs, and new Abnormal User Plane Traffic and traffic pattern analytics and related VFL measurements were added.
- “The NWDAF can provide analytics output and optionally analytics accuracy value or loss value to the analytics consumer(s)” TS 23.288 v20.2.0
- “The NWDAF shall be able to provide the statistics and predictions on Abnormal User Plane Traffic Analytics to consumer NF” TS 23.288 v20.2.0
- Add analytics and ML model accuracy checking capability information for NWDAF TS 28.541CR1714
Defining Specifications
3GPP specifications that define or reference NWDAF, with the latest known release. Sourced from the 3GPP document catalog — see methodology.
| Specification | Title | Release |
|---|---|---|
| TS 23.435 vk00 | Network Slice Capability Enablement (NSCE) Architecture | Rel-20 read |
| TS 23.436 vk20 | Application Data Analytics Enablement | Rel-20 read |
| TS 23.482 vk30 | AI/ML Enablement for Verticals | Rel-20 read |
| TS 23.501 vk30 | 5G System Architecture | Rel-20 read |
| TS 23.503 vk30 | 5G Policy and Charging Control Framework | Rel-20 read |
| TS 23.700 vk10 | AI/ML Application Layer Support Phase 2 | Rel-20 read |
| TR 23.758 vh00 | Study on Edge Application Architecture | Rel-17 read |
| TS 23.791 vg20 | NWDAF Data Collection & Analytics Framework | Rel-16 read |
| TS 26.531 vj00 | Data Collection & Reporting Architecture for 5G | Rel-19 read |
| TS 26.532 vj10 | Data Collection and Reporting API Specification | Rel-19 read |
| TS 26.804 vk10 | 5G Media Delivery Architecture Extensions | Rel-20 read |
| TR 26.812 vi10 | Technical Report | Rel-18 read |
| TS 26.941 vk00 | 5GMS Network Slicing Study | Rel-20 read |
| TS 26.942 vk00 | Sustainable Media Metrics and Architectural Impacts for 5G | Rel-20 read |
| TS 28.201 vj20 | 5G Network Slice Performance Analytics Charging | Rel-19 read |
| TS 28.535 vj00 | Closed Control Loop Assurance Management | Rel-19 read |
| TS 28.536 vj20 | Management services for communication service assurance | Rel-19 read |
| TS 28.879 vj10 | OAM for Service Management Exposure Study | Rel-19 read |
| TS 29.503 vk10 | UDM Service Based Interface | Rel-20 read |
| TS 29.507 vk10 | Access and Mobility Policy Control | Rel-20 read |
| TS 29.508 vk10 | Session Management Event Exposure Service | Rel-20 read |
| TS 29.510 vk10 | NRF Service-Based Interface | Rel-20 read |
| TS 29.512 vk10 | Session Management Policy Control Service Stage 3 | Rel-20 read |
| TS 29.513 vk10 | 5G PCC Signalling Flows | Rel-20 read |
| TS 29.514 vk10 | Policy Authorization Service | Rel-20 read |
| TS 29.517 vk10 | AF Event Exposure Service | Rel-20 read |
| TS 29.518 vk10 | AMF Service Based Interface | Rel-20 read |
| TS 29.520 vk10 | 5G NWDAF Services | Rel-20 read |
| TS 29.521 vk10 | Binding Support Management Service | Rel-20 read |
| TS 29.523 vk10 | Policy Control Event Exposure Service | Rel-20 read |
| TS 29.530 vk10 | 5G AF AI/ML Services | Rel-20 read |
| TS 29.536 vk10 | Nnsacf Network Slice Admission Control | Rel-20 read |
| TS 29.543 vk10 | Data Transfer Policy Control Services | Rel-20 read |
| TS 29.551 vk00 | 3GPP TS 29551 vk00: PFD Management Service Specification | Rel-20 read |
| TS 29.552 vk10 | 5G Network Data Analytics Procedures | Rel-20 read |
| TS 29.554 vk10 | BDT Policy Control Service | Rel-20 read |
| TS 29.562 vk10 | HSS Nhss Services for IMS and GBA | Rel-20 read |
| TS 29.564 vk10 | Nupf Service Based Interface | Rel-20 read |
| TS 29.574 vk10 | DCCF Service Based Interface | Rel-20 read |
| TS 29.575 vk10 | ADRF Service Based Interface | Rel-20 read |
| TS 29.576 vk10 | MFAF Service Based Interface | Rel-20 read |
| TS 29.591 vk10 | NEF Southbound Services | Rel-20 read |
| TS 29.889 vj10 | Study on UPF data collection for AI/ML | Rel-19 read |
| TS 29.890 vg00 | CT3 5G System Technical Report | Rel-16 read |
| TS 32.240 vk10 | 3GPP Charging Architecture and Principles | Rel-20 read |
| TR 32.847 vi00 | Technical Report | Rel-18 read |
| TS 33.127 vj80 | Lawful Interception Architecture | Rel-19 read |
| TS 33.794 vj10 | Study on Zero Trust Security Enablers for 5G | Rel-19 read |
| TR 33.866 vh00 | Security aspects of Network Automation enablers for 5GS | Rel-17 read |
| TR 33.867 vh10 | User Consent for 3GPP Services | Rel-17 read |