5G Demonstration on an EMDC with ML-Enabled Scaling and QKD-Secured Connectivity

This paper presents the implementation of a fully functional 5G network on a compact, energy-efficient edge micro datacenter, showcasing features such as workload prediction and preemptive scaling. The network extends across multiple edge micro datacenters and incorporates quantum-secure connectivity. Our demonstration provides a blueprint for forward-looking 5G deployments aiming to meet challenging latency and throughput requirements while complying with stringent security requirements. Measurements are performed for network throughput and latency as well as for CPU load of the different components of the 5G deployment, including distributed unit, centralised unit control and user planes, 5G core and the RAN intelligent controller. Additionally, an intelligent workload prediction mechanism, based on an LSTM model, enables preemptive scaling of the centralised unit's user plane. This proactive approach helps to mitigate bottlenecks as the number of users connecting to the network increases.