Slice-Level Performance Metric Forecasting in Intelligent Transportation Systems and the Internet of Vehicles

Abstract

The intricate web of vehicles connected in the fifth-generation (5G) wireless infrastructure forms the Internet of Vehicles (IoV) and enabling technologies, such as Multi-Access Edge Computing (MEC) and network slicing, are employed in guaranteeing application requirements in the IoV and optimizing network resource allocation. In particular, network slicing allows mobile network operators to support virtualized end-to-end networks with diverse slice requirements that are typically grouped into use case classes such as Enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communication (uRLLC), and Massive Machine-Type Communication (mMTC). These enabling technologies rely on Performance Metrics, monitored and gathered within the network, in order to evaluate and suggest improvement for slice configuration. As such, this article considers a forecasting model for Performance Metrics at the network slice level by leveraging the use of the Network Data Analytics Function (NWDAF) and its edge placements. The results and analysis, including the scalability of the forecasting model, are assessed as a step towards total automation of network slice management within the 5G network. This evaluation is later illustrated using an end-to-end IoV use case incorporating the edge NWDAF placements to guide decision-making regarding management and orchestration for future networks.