NWDAF: The Gateway to AI and 5G Monetisation
mayo 12, 2021 - Sara Philpott
With 5G there will be an explosion of available data. A new function will manage and monetise that new data.
With 5G, the Network Data and Analytics Function (NWDAF) is a new core function that will introduce a higher level of intelligence to the network. This will make 5G networks smarter through real-time data management and analysis which will drive a range of systems, actions and decisions in a CSP’s business.
NWDAF, with combined data abstraction and machine learning (ML) modelling and inference based KPIs per use case, provides essential input for:
- Closed loop automation (analytics, AI and policy) for service performance and network optimisation essential for guaranteeing quality of experience (as well as productivity savings)
- Insights though auto-discovery that are essential for monetization, new service and offer innovation
The exciting aspect from a 5G perspective is that these can be done on a network slice and per service level, providing a finer granularity of focus for insight.
Analysts have agreed for some time that 5G monetisation depends on Telco Cloud success and that NWDAF provides a key component in the path towards autonomous networks. NWDAF monitors all aspects of each network ‘slice’ for compliance with relevant service-level agreements. Operators can make use of NWDAF in their infrastructures to help control service level and deliver strong customer experiences for a diverse range of new services.
The analyst Analysys Mason stated in November 2020, that there is “a clear need for the NWDAF in order to standardise operational data for all applications and functions… Over 100 carriers worldwide are implementing 5G networks using 3GPP standards. 5G networks are already complex, and this complexity will increase as core network components that enable network slicing are added. A high degree of automation is therefore required to avoid increasing operational costs. AI and analytics solutions are now seen as critical for the creation of new automations; this was first recognised in the 3GPP standards in 2017 when the NWDAF standard was initially defined.”
In other words: standards are fine but it will need more than standards for true differentiation.
Automation introduces a higher level of intelligence in the network, new learning architectures for multi-vendor deployments, new concepts for the RAN, core and OSS to further optimise the performance and efficiency.
Adjacent functions rapidly come into play. A key part of managing that automation is closed loop use cases – using the scoring and inference information derived from AI models in order to initiate an automatic policy (or other) action in response. Policy actions are determined by the rules governing the decisions to be made. By leveraging analytics, self-learning algorithms and policy, closed loop use cases become adaptive and responsive to resource optimisation and performance sensitivities.
The 3GPP standards prescribe a number of use cases for NWDAF (9 in Release 16, with additional enhancements and use cases under review in Release 17), directed at automating Network Operations in order to deliver cost and performance efficiencies. Beyond these standard use cases we see the opportunity to combine, correlate and apply algorithms to the insight gained from network sources with the BSS transaction activity in order to produce similar efficiencies for Service Operations and Monetisation.
Amdocs believes that the NWDAF will therefore provide the necessary platform and gateway for further use case introduction (license and service growth) as well as a gateway for AI and 5G Monetisation (enabling autonomous networks and auto-discovery).
A paradigm shift in productivity, service enhancement and optimisation will be experienced with the adoption of autonomous networks and AI. We believe NWDAF will be central to that, along with opening the doors to new AI driven 5G monetisation.
For more on Openet’s work in helping service providers to better manage their 5G data, visit Openet Data.
BLOG
