Performance Improvements in Software-Defined and Virtualized Wireless Networks

Abstract

In this dissertation, we investigate performance improvements in software-defined and virtualized wireless networks with advanced in-network caching and mobile edge computing (MEC). Wireless network virtualization (WNV), software-defined networking (SDN), MEC and in-network caching are new promising technologies in next generation wireless networks. Traditionally, in-network caching and MEC have been addressed separately from WNV and SDN. However, it is necessary to jointly consider in-network caching and MEC with WNV and SDN together to provide better services in future wireless networks. Therefore, in this dissertation, we propose to integrate WNV and SDN with in-network caching and MEC in order to improve the end-to-end network performance in wireless networks. We firstly show that jointly considering WNV and in-network caching is necessary and develop an efficient alternating direction method of multipliers (ADMM)-based distributed virtual resource allocation and in-network caching scheme. Secondly, motivated by the experience of user equipment admission control in traditional wireless networks, we propose a novel concept of virtual network (VN) admission control for wireless virtualization. By limiting the number of VNs embedded in the physical network, VN admission control can effectively guarantee the quality of service experienced by users of VNs and maximize the utilization of the physical networks at the same time. Thirdly, we propose to jointly optimize video streaming, bandwidth provisioning and caching strategies in software-defined wireless networks (SDWNs) with limited network resources and quality of service (QoE) requirements. We design a novel mechanism to jointly provide proactive caching, bandwidth provisioning and adaptive video streaming. Lastly, in addition to caching, we integrate MEC into the considered SDWN to enhance the video service in next generation wireless networks. We utilize dual-decomposition method and ADMM design a decentralized algorithm to solve the proposed problems. Simulation results are presented to show the effectiveness of the proposed schemes.