Artificial intelligence (AI) is leading to an automation revolution in 5G, beyond 5G, and 6G cellular networks. AI is also empowering the trend to open up the 5G/6G network capabilities by leveraging universal infrastructure, open network architectures, open-source software/hardware, and other state-of-the-art technologies (such as software-defined networking, network function virtualisation, multi-access edge computing, network slicing, and artificial intelligence, etc.). AI can impact all aspects of cellular network design from the physical layer waveform to radio resource management to the network core to the applications that make use of the network.
The 5th generation of mobile (and fixed) devices, commonly known as 5G, is being standardized by 3GPP. 3GPP groups a set of standards into Releases, which include a collection of functional packages of technologies. The first round of 5G standards, known as Release 15, completed its standardization at the end of 2018. Compared to previous releases of the specifications, Release 15 5G Core included a new network function known as Network Data Analytics Function (NWDAF) [3GPP, “Architecture enhancements for 5G System (5GS) to support network data analytics services,” 3rd Generation Partnership Project (3GPP), Technical Specification (TS) 23.288, December 2021, version 17.3.0. [Online]. Available: https://www.3gpp.org/DynaReport/23288.htm], whose main purpose is to aid the network in issuing analytic reports that can be used for assisting and influencing other 5G Core network functions in making automated decisions. While in Release 15 the use of NWDAF was limited to a single-use case, throughout Release 16 (whose standardization was completed at the end of 2020) more use cases were added, which provided additional functionalities for easing the deployment, administration, of management of the network automation functionality. This paper provides an overview of the use cases, architecture, and functionality of the Network Data Analytics Function (NWDAF) in 5G Core networks, taking into account also the architectural work pertaining to Release 17 which has just concluded and added a few additional features and use cases.
The need for network automation guided the design of the 3GPP system standardized in Rel. 15. Since this release the generation of data (and analytics) from the network is not relegated anymore to metrics exchanged between the network element and their element manager, almost always using proprietary interfaces; instead, the system architecture has been re-designed to natively support automation loops. The keystone of this system is the Network Data Analytics Function (NWDAF), which gathers data, computes analytics, and shares them with other functions in the system. In Release 16, 9 analytics were defined, as depicted next.
The 5G Core functions act as both producers and consumers for the different analytics. This ecosystem has been further extended for Release 17 and Release 18, adding more analytics and re-structuring the analytics hub.