Evaluating Software-Defined Networking-Driven Edge Clouds for 5G Critical Communications

摘要

Modern societies are increasingly dependent on Critical Infrastructures (CIs) such as Smart Grids (SGs) and Intelligent Transportation Systems (ITS). Due to continuously rising demands in terms of efficiency and capabilities, complex control and monitoring strategies are employed. These in turn rely on robust, high performance communication networks. As CIs are distributed systems with diverging demands, the deployment of individual networks is often too costly and time consuming. Furthermore, the vast physical scope traditionally results in unacceptably high end-to-end delays. Therefore, a convergence of public and dedicated networks, as well as traditional Information and Communication Technology (ICT) with Information Technology (IT), are considered key aspects of 5G for addressing these challenges. Hence, this paper provides an empirical evaluation of CI communication services on basis of a Software-Defined Networking (SDN) and Network Function Virtualization (NFV) driven Edge Clouds (ECs) within a sliced 5G network. By shifting computing resources from the backbone or back office towards the network's access level, ECs allow for drastically reduced delays. Also, the traffic load on several layers of the communication infrastructure is reduced, as data can be kept locally, i.e. close to the source. This is demonstrated by shifting an ITS application from a central cloud to the EC. Services are transferred step-wise and transparently, minimizing interruptions while dynamically adapting to the backhaul's available data rate. The developed system is evaluated under realistic traffic conditions within a physical testing environment.