Orchestration and Performance Evaluation of 5G-based Drone Applications

Abstract

5G has seen unprecedented growth in recent times. Its features are paving the way for new opportunities, and Network slicing is one of them. Slicing enables virtualization across the entire provider network architecture to create on-demand network slices that meet service requirements. Because of network slices, new verticals are deployed on top of 5G networks. One such vertical that is gaining popularity because of its use cases is 5G drones. The 5G-based drone application runs on a container orchestration platform while the UAV is connected to the 5G network. Remote management of UAV applications running on top of a 5G network is challenging. Monitoring KPIs related to the cloud and 5G network in near real-time will facilitate the management of UAV applications running in a network slice. In this thesis, we deployed a monitoring stack comprising Prometheus, Zabbix, and ELK stack to monitor, store and visualize KPIs. Prometheus collects metrics from an application deployed on Kubernetes, while Zabbix gathers metrics from the 5G network parallelly in near-real time. The monitoring stack successfully gathers and visualizes the KPIs of a 5G-based drone application for remote management. The deployed monitoring stack will provide insight to a service provider and drone players.