Network Data Analytics Function in 5G: Automated Analytics Where You Need It

It’s no secret that companies like Google, Amazon, Facebook, and Apple have built their success on customer-experience innovations. They have higher net promoter scores (NPS) and satisfaction rates because they know what customers want even before customers do. This sets the bar high for what people and companies expect from all their providers in terms of digital engagement and digital experiences.

Higher customer expectations will be a huge component of 5G services, especially those with service guarantees (as in network slicing, IoT-driven autonomous and smart services, fixed wireless access, enhanced mobile broadband, to name a few).

To effectively unlock the potential of 5G services, operators and enterprises alike will have to understand what customers are experiencing in real time. But with analytics woven into 5G call flows, there’s the danger of “too much data, too little insight,” especially if there is a deluge of data across core, transport, and edge domains. The data volume is expected to grow as the dimensions that define customer experience also grow, through mashups, multiplexing, features, devices, locations, OTT services, and network topologies. This is a phenomena we examine closely in our Global Internet Phenomena Report's “Growing Network Complexity” spotlight.

To cut through that complexity, predictive analytics platforms must be sophisticated enough to examine data from 5G core networks and fuel insight-driven decision making. According to ABI Research’s recent report “5G Network Analytics and NWDAF, this is a market that will grow to US$6.4 billion by 2026. Currently, however, less than 2% of data generated has been stored, and within that, less than 10% has been analyzed and applied, according to ABI.

The problem is that for too long, service providers have had to integrate proprietary solutions – each with their own sets of KPIs and characteristics – for overlays of stats and analytics. A true service-assurance approach to 5G will require analytics throughout every domain of the network.

Enter Sandvine Network Data Analytics Function (NWDAF)

In order to collect data from user equipment, network functions and operations, administration, and maintenance (OAM) systems, the 3GPP’s Network Data Analytics Function (NWDAF) extracts data from microservice functions residing inside the network core so that relevant data can be analyzed at scale. Though the NWDAF is designed primarily for core network and OAM systems, we go a step further by capitalizing on the fact analytics functions sit in other domains of the network – like the RAN layer, and the app function layer – where KPIs and metadata are collected.

Because of our 5G Standalone (SA) expertise, we are in a unique position to take the NWDAF analytics service to the next level. In Sandvine’s 3GPP-compliant NWDAF Service Intelligence Engine, the core can inform the end-to-end network components about what’s happening through the DAF via a messaging bus. 3GPP Release 15 makes this possible by integrating NWDAF with messaging in the service-based interface architecture (SBI). Because the NWDAF is designed to be very open, any device and any network function can go to the bus and contact any network functions (NFs). So, though the NWDAF is dedicated to the core, the Sandvine Service Intelligence Engine enables cross-domain communication, which ultimately helps improve application quality of experience.

Below are five differentiating factors for Sandvine’s NWDAF, which uniquely visualizes actual network conditions so that operators can improve end-to-end app quality-of-experience:

1. Advanced application classification offers fine-grained analysis, which is coupled with real-time visualization of trends at 250ms intervals to reveal network/performance problems (packet loss, throughput, and latency) and revenue-generation opportunities (best-performing apps). By ingesting Application and Network Intelligence (ANI) data in addition to that about user equipment (UE), Network Function (NF), Analytics Function (AF), and Operations and Management (OAM) data, the Service Intelligence Engine becomes a robust source for many of the KPIs and metadata operators need to collect from different vendors’ NFs and OAMs;
   
   
2. Enriched Outputs provide an option to take in highly granular contextual user plane KPI data, which enriches analytics services and use cases with user and application QoE awareness. Observed Service Experience Analytics ID with Application and Network Intelligence (ANI) provide a single source of metrics and KPIs for the analytics service (that means no dependency on AF);
   
   
3. Sophisticated visualization of data with Analytics IDs presents information with out-of-the-box intuitive use case dashboards that can also be exported to third-party systems for planning and operations teams’ consumption. Use cases include Network Slice SLA Assurance; UE Policy Optimization; Congestion Management; QoS Assurance for V2X/D2X; and Cyber Threat Analysis and Management;
   
   
4. Leverages 3GPP’s User Plane Function (UPF) Load Analytics with an NF load analytics service that offers real-time operational intelligence in the form of statistics and predictions about different users/consumers in the 5G Core network, including NFs and OAM services. Because the Service Engine incorporates an average of user-plane data scores (a composite KPI calculated using more fundamental network KPIs like bandwidth, latency, and packet loss), it better assures the delivery of high-speed and low-latency services.
   
   
5. Accurate prediction of network, slice, and user equipment (UE) performance across the entire network, while also precisely understanding and managing each application flow and its network path to meet performance and service level agreement (SLA) needs. Coupled with real-time policy actions, the NWDAF’s predictive insights become actionable.

These NWDAF enhancements ensure Sandvine’s enterprise and service provider customers can ask any device for stats and predictions about any device in or out of the network. The highly granular insight enhances network automation and service orchestration efforts, helping our customers adjust to the requirements imposed by the service. They can then optimize network resources and better assure service QoE.

As our customers transform deployments to be more containerized and cloud based for emerging cloud and slicing services, Sandvine’s Service Engine will enable an application QoE-driven approach to assurance and monetization of 5G services for enterprises, IoT, and consumer customers.

To learn more, check out our Service Engine and NWDAF resources page; our ABI custom report “5G Network Analytics and NWDAF”; and our white paper “5G Service Intelligence Engine for Core, Cloud, and Edge Networks.”

Please contact us with any questions.