Towards 5G: An Empirical Evaluation of Software-Defined End-to-End Network Slicing

Emerging systems such as the Internet of Things (IoT), Smart Grids (SGs), Industry 4.0, and Vehicle-to-Everything (V2X) impose a diverging set of requirements on underlying Information and Communication Technology (ICT) topologies. This relates to performance indicators including network delay, data rate, reliability as well as security. To meet these criteria, 5G aims to provide network slices, i.e., virtually independent architectures on top of a single, unified communication infrastructure. In this context, Software-Defined Networking (SDN) and Network Function Virtualization (NFV) have been identified as key technologies for implementing such a solution. This work utilizes these ingredients in conjunction for providing isolated end-to-end slices, from wireline cloud servers over the Software-Defined Radio (SDR)-based Long Term Evolution (LTE) Radio Access Network (RAN) down to individual User Equipments (UEs). Thereby, a prototypical 5G end-to-end slicing solution is designed and implemented. The system enables independent management of each slice by either its respective owner or the ICT infrastructure operator. Also, resources are allocated dynamically to slice tenants in order of their priority. A comprehensive empirical evaluation based on real-world traffic patterns, such as Floating Car Data (FCD), is given. Hence, compliance with Service Level Agreements (SLAs) - crucial to stable operation of e.g., SGs - is demonstrated for up- and downlink traffic flows.