Title: An NWDAF Approach to 5G Core Network Signaling Traffic: Analysis and Characterization

Abstract: Data-driven approaches and paradigms have become promising solutions to efficient network performances through optimization. These approaches focus on state-of-the-art machine learning techniques that can address the needs of 5G networks and the networks of tomorrow, such as proactive load balancing. In contrast to model-based approaches, data-driven approaches do not need accurate models to tackle the target problem, and their associated architectures provide a flexibility of available system parameters that improve the feasibility of learning-based algorithms in mobile wireless networks. The work presented in this paper focuses on demonstrating a working system prototype of the 5G Core (5GC) network and the Network Data Analytics Function (NWDAF) used to bring the benefits of data-driven techniques to fruition. Analyses of the network-generated data explore core intra-network interactions through unsupervised learning, clustering, and evaluate these results as insights for future opportunities and works.