Competing in 5G requires telecom operators to move beyond consumers as their target audience and attract enterprise customers with increasingly enhanced experiences and custom-made services for different market verticals. In addition, operators have already made significant investments in rolling out 5G. To achieve a solid return on investment, operators must identify and develop new differentiated services and support new B2B2X business models.
Furthermore, 5G is a purpose-built network designed to facilitate an always-connected world with use cases like the next industrial revolution, known as Industry 4.0. Applications ranging from industrial automation, robotics, AR/VR, telemedicine, and immersive experiences will all be enabled by enhanced 5G networks.
5G is already much more complex than previous network generations, with more than 400 network procedures in 5G networks, each with dedicated key performance indicators (KPIs) and processing algorithms. Yet, 5G places strict network efficiency and reliability requirements to deliver such advanced 5G services as industrial automation. The complexity is further intensified when considering that there will be 100 times more connected endpoints and a heterogeneous network topology driven by Mobile-Access Edge Computing (MEC), enterprise services, and 5G private networks. Also, there will be IoT use cases with time-sensitive applications that require dynamic network resource allocation.
This requires operators to monitor their networks in real-time and run them using automated systems. How can operators achieve these business objectives? How can they identify new opportunities and deliver on their promise? How can they manage this complexity while ensuring efficiency and differentiating themselves?
So, what is the solution?
In this article, we will explore the 3GPP-defined Network Data Analytics Function (NWDAF) and how operators can adopt a platform-based approach to analytics as a built-in network function, which will provide a standardized, company-wide solution to data analytics and AI/ML to help manage 5G complexity, improve the customer experience, reduce costs, and generate revenue streams, all while making engineering teams more efficient.
A data analytics function acts as the operator’s conductor to direct the network’s performance to ensure that all functions are working in harmony and that the service output is of the highest quality.
At present, less than 2% of data generated is stored. Within that 2%, less than 10% of the data stored is analyzed and utilized to drive decision-making and gain insights. In other words, 98% of data generated is not being used.” This is significant. Data is the fuel that propels businesses forward and helps them make informed decisions and gain strategic insights. This sentiment is summed up nicely in a quote by Peter Sondergaard, Senior Vice President and Global Head of Research at Gartner, Inc.
Information is the oil of the 21st century, and analytics is the combustion engine.
Large-scale cloud companies (like Google, Amazon, and Meta) have been leveraging analytics to make better data-driven decisions for over a decade. By contrast, in the telco sector, hundreds of operators innovate and roll out cutting-edge technology yet need to leverage analytics and AI/ML. In the era of 5G, they are flush with data riches. This fuel can be fed into the AI/ML algorithms combustion engines to generate insights to save costs, make operations more efficient, and drive network automation.
So, how can operators efficiently deploy data analytics into their networks?
To help operators adopt a unified approach to data analytics, the 3rd Generation Partnership Project (3GPP), the standards body developing mobile telecommunications protocols, introduced the Network Data Analytics Function (NWDAF) framework as part of the 5G standard.
Specified as part of the 5G Service-Based Architecture (SBA), NWDAF provides a broad and deep set of analytics to drive actionable insights. It offers real-time operational intelligence to request NFs for network automation, service orchestration, and operational events. NWDAF consumes KPIs from NFs, processes this stream of information in real-time, and provides predictions to NFs and other systems via subscriptions and pull interfaces.
The 5G-PPP Architecture Working Group View on 5G Architecture Version 4.0 Â and the End-to-End Data Analytics Framework for 5G Architecture both describe multiple data analytics functions (DAFs), Management Data Analytics Function (MDAF, which can act as a domain-specific function or cross-domain), Application Function Data Analytics Function (AF-DAF), Radio Access Network Data Analytics Function (RAN-DAF), and Data Network Analytics Function (DN-DAF).
Some vendors create a unique solution for each area of the network. However, using the same cloud instance, RADCOM’s underlying architecture for the NWDAF can be configured to serve as a different data analytics function, such as RAN-DAF and MDAF. Here is the basic functionality of the NWDAF and its architecture:Â Â
By adopting this standardized analytics framework, operators can implement a unified approach to analytics across the organization and transform their operations by placing data and analytics at the heart of their business. The function also provides predictive analytics as it learns and sets baselines for service quality, traffic loads, and network usage, enabling operators to understand how fluctuations in the network and subscriber behavior impact the business.Â
NWDAF provides multiple use cases that will expand in Release 18: PFD determination analytics, location accuracy analytics, end-to-end data volume transfer time analytics, relative proximity analytics, PDU Session traffic analytics, and movement behavior analytics. Here are a few current examples split into three broad categories:
In February 2023, RADCOM announced that it was partnering with Rakuten Mobile on the telecom industry’s first Network Data Analytics Function (NWDAF) project. Our continued investment in developing a native network function dedicated to data analytics has proven its value. Here are some of the use cases, split into three different categories:
NWDAF takes assurance to the next level by monitoring and automating processes to ensure network performance and service quality while making engineering teams more efficient and able to prioritize tasks with the help of AI.Â
The NWDAF uses predictive analytics to identify the load level in each network slice instance and creates slice utilization KPIs provided to 3rd party NFs, such as the PCF (Policy Control Function) and NSSF (Network Slice Selection Function) per network slice instance. It constantly monitors each network slice and predicts the quality of experience. If, based on forecasting, the NWDAF detects a downward trend due to congestion and that the QoE score will drop below the SLA, the NWDAF alerts the NSSF so that the slice selection policy changes so that all new subscribers connect to an alternative slice.
The NWDAF detects a signaling storm by monitoring signaling traffic rates in the Control Plane NFs such as AMFs, SMFs, etc. Built-in AI/ML is applied to the signaling traffic rate data to automatically detect and isolate the root cause and UEs and establish which gNodeBs generate the signaling storm. As the NWDAF sees a storm is starting and before the storm severely affects the NFs, the NWDAF triggers corrective actions to prevent the storm, such as disabling gNodeB links, temporarily barring UEs or reconnecting UEs to other gNodeB/AMFs.
The NWDAF has been estimated to prevent at least 75% of signaling storms. This is hugely significant and already means saving millions in costs to the operator and ensuring a more stable network for customers. In addition, further model training can improve this number by training the AI/ML model and feeding it more data.
NWDAF uses its built-in AI capabilities and ability to trigger closed-loop automation to reduce costs and make the network more efficient.Â
In a blog post, Nokia reported more than 28% of the total signaling load is paging. This is a significant load on the network, which can be optimized using NWDAF. We reduced paging by at least 46% compared to traditional paging methods based on the last X cell or defaulting to the closest cell geographically.
In the screenshot above, the journey of the UE shows past locations with a yellow triangle. The green triangle indicates the UE’s current location (diagram on the left). In traditional paging, the UE’s last X cells are paged. However, the NWDAF uses predictive analytics to estimate which cells are most likely in the UE’s path.Â
In this example, the NWDAF gave its best estimate to the cell on the left of the UE’s current position (with a 40% probability), and this was the cell closest to the UE as it moved on its journey (diagram on the right).   Â
Utilizing the NWDAF for optimization significantly reduces costs and makes the network more efficient. The AMF uses analytics from the NWDAF, which predicts the UE location in real-time, allowing the AMF to page fewer cells and reduce the use of network resources and air interfaces, achieving a significant reduction in energy consumption as well as a reduction in mobile terminating call set-up time.
The operator can utilize NWDAF to optimize the network further, generate revenues, save costs with internal monetization, and drive revenues with external monetization by selling data analytics-as-a-service to multiple market verticals.
Enhanced AI analytics enables operators to extract insights from network data to improve efficiency and reduce costs. For example, insights about traffic patterns, network load, and other factors can be leveraged to optimize power consumption and reduce energy-related costs. Furthermore, by tracking and identifying the level of connectivity in and around a city or even in rural areas, operators can see where additional sites and infrastructure are needed for better planning that optimizes costs.Â
Moreover, using AI/ML-based analytics, operators can use churn prediction, automatically detecting customers likely to cancel their subscriptions and acting in time to prevent them from doing so.Â
NWDAF analytics can also enable telecom operators to provide real-time contextual offers or drive contextual advertising. For example, operators can upsell data to subscribers who are underutilizing their quota or push content for those only in areas with ample available bandwidth.
Operators can extract insights from data about user behavior, movement, commute patterns, and more to sell them to entities across multiple industries and grow revenues. For example, location-based data can be sold to advertisers and retailers. Insights can also enable a local municipality to understand which routes are most popular among commuters and make better data-driven decisions about whether a new bridge or a highway renovation project would better serve the public. Also, network data insights can be sold to transportation service providers to make more informed decisions about which lines and routes need to be added or augmented so they can meet demand and grow revenues. These data insights can also be sold to video content and application providers for use cases like personalized or location-based advertising.Â
Operators have a unique opportunity to use analytics to create new charging strategies by utilizing NWDAF’s integration with the charging functions in 5G. These can be implemented by the NWDAF and the Charging Enablement Function (CEF) interacting with the Charging Function (CHF).
Due to its cloud-native architecture, the NWDAF function is modular and agile, allowing for various implementation models that can be on-prem/off-prem and centralized or decentralized. One of the options is a centralized/NWDAF proxy. One of the use cases for this deployment model is in private networks where enterprise customers require analytics with a low footprint. A lightweight NWDAF is installed on a private network on-prem and collects the raw data from the network. The aggregated data is sent to a centralized NWDAF. The centralized DAF computes the analytics per enterprise network and sends the analytics output to the lightweight NWDAF instance on-prem. This proxy then triggers corrective action within the private network.Â
Another use case for the centralized/NWDAF proxy model is that large-scale operators with multiple cloud networks can deploy lightweight proxy NWDAFs and utilize the centralized NWDAF to perform network-wide analytics. So, for example, operators can analyze UE traffic trends across their entire network.Â
NWDAF can help empower operators to optimize network performance and deliver superior customer experiences, all while saving costs. From automated assurance, network optimization, and predictive analytics, NWDAF enables telecom operators to stay ahead in a rapidly evolving 5G landscape. RADCOM is an operator’s 5G co-pilot that brings enhanced, telco-specific AI analytics to help transform network operations, reduce time to resolution, and make teams more efficient.
If you want to learn more about how RADCOM can help you take a unified approach to 5G data analytics, visit https://radcom.com/products/products-radcom-nwdaf/Â
Subscribe for industry insights. Elevate your influence – promote with us!
