RESOURCE ALLOCATION ALGORITHMS FOR UPF INSTANCES IN THE 5G CORE NETWORK

Abstract

The service based architecture of 5G core network allows network services to flexibly and automatically scale up and down according to user traffic. Cloud-based implementation of 5G core network functions is a promising approach for serving increasing 5G user devices. However, allocating cloud resources for scaling 5G core network functions is an NP-hard problem due to the diversity of user demands and multiple criteria from both service providers and users’ requirements of service quality. At present, the deployment of 5G network function on cloud is still limited. It is crucial to investigate the cloud-based approach for ensure full deployment effective and efficient in the near future. This study proposed two algorithms, Load-Balance and Energy-Saving, for cloud resource allocation to automatically scale up and down the 5G UPF instances. A simulation software is developed to model an Infrastructure-as-a-Service cloud and implement these resource allocation algorithms. Numerical results indicate that Energy-Saving can lead to better resource utilization and reduce over 33% energy consumption, while Load-Balance assures cloud server to not overloaded.