Experimental evaluation of a Follow-me MEC Cloud-Native 5G network

Abstract

Cloud-native approaches for network functions In the 5G context have been embraced by the community, as they allow flexible management, reconfiguration and monitoring of the network in an end-to-end manner. As network softwarization extends to the RAN, empowered through the cellular stack disaggregation, even base station components can be executed as a cloud-native service. As low latency access is needed in the 5G and beyond networks for serving ultra Reliable Low Latency Communications, Multi-Access Edge Computing (MEC) needs to be integrated in the overall architecture. As the user moves among different RANs, the latency of accessing the service needs to be preserved for providing users with a seamless experience. To accomplish such behavior, migrations of the hosted services are needed, though not fully compatible with the cloud-native approach, and placing them closer to the network access point of the user. In this work, we experiment with a cloud-native end-to-end network, enhanced with Follow-me MEC functionalities. Heterogeneous access is provided at the RAN level, using disaggregated base stations, and MEC is integrated on the fronthaul of the network, ensuring low-latency access to services. The entire network is instantiated in a cloud-native manner, using a widely adopted container orchestration solution. Our results show that the scheme is able to provide low latency access to the hosted services, while the UE remains agnostic of the entire process and without any drops of the already established connections.