Data Collection Using NWDAF Network Function in a 5G Core Network with Real Traffic

5G aims at providing significant improvements to mobile networks with regard to previous 4G ones, in both performance and flexibility. Indeed, a 5G network is expected to increase throughput and reduce latency, leveraging softwarization and virtualization concepts and the state of the art in radio access and resources allocation. A 5G system (5GS) long-term goal is to address more and more challenging and broader use-cases, allowing a multi-tenant use (slicing) of the same physical infrastructure to enable and deploy at the same time ultra-reliable and low latency, machine-type or broadband communication services. In that context, network resources are highly abstracted and require complex management operations, which ultimately must be reflected downwards to the physical resources. In supporting that task, since 3GPP Rel-16, a network function called Network Data Analytics Function (NWDAF) was introduced, with the primary goal of collecting and processing heterogeneous data coming from the User Plane of the 5GS. The NWDAF outputs can then be used to perform network reconfigurations and automation in general, optimizing traffic management and overall network performance. In this paper we present which realtime metrics can be collected by a NWDAF and how, in a reference 5G core network, validating the correctness of the measurements performed using real traffic in a Linux-based testbed.