A multi-stage analysis of network slicing architecture for 5G mobile networks

Abstract

Recently, the increasing demand for low latency, the explosive growth in the volume of network traffic, the large and growing number of connected devices, and diversified multimedia applications have paved the way for a new era of mobile networks. To meet these diverse requirements of different businesses in network virtualization, network slicing has emerged as a promising paradigm of upcoming 5G mobile networks. Network slicing is a major technology, based on network function virtualization and software defined network technologies, which aims to achieve more efficient utilization of available network traffic and reduce operating costs. In this paper, we propose a network slicing architecture for 5G mobile networks involving cloud radio access network (C-RAN), mobile edge computing (MEC), and cloud data center. We model the proposed network slicing system based on queueing theory, which can be used to derive the main performance metrics such as the CPU utilization, system throughput, system drop rate, average number of message requests, average response time, and average waiting time. We provide quantitative examples to show how this proposed model could be applied to estimate the system performance and cost for a network slicing system in 5G mobile networks and the number of C-RAN and MEC cores required under diverse 5G traffic conditions. The analytical results and simulation models indicate that the proposed model has a powerful ability to assign the number of C-RAN and MEC cores required to achieve the quality of service targets of 5G slices.