Estimation of 5G Core and RAN End-to-End Delay through Gaussian Mixture Models

Abstract

Network analytics provide a comprehensive picture of the network’s Quality of Service (QoS), including the End-to-End (E2E) delay. In this paper, we characterize the Core and the Radio Access Network (RAN) E2E delay of 5G networks with the Standalone (SA) and Non-Standalone (NSA) topologies when a single known Probability Density Function (PDF) is not suitable to model its distribution. To this end, multiple PDFs, denominated as components, are combined in a Gaussian Mixture Model (GMM) to represent the distribution of the E2E delay. The accuracy and computation time of the GMM is evaluated for a different number of components and a number of samples. The results presented in the paper are based on a dataset of E2E delay values sampled from both SA and NSA 5G networks. Finally, we show that the GMM can be adopted to estimate a high diversity of E2E delay patterns found in 5G networks and its computation time can be adequate for a large range of applications.