The target of Network Function Virtualization (NFV) is to decouple network hardware and software, enable the development of network solutions by using multi-vendor hardware that can effectively cooperate with multi-vendor software platforms and replace current standard networking operations for fixed and mobile networks. It is hoped that CAPEX and OPEX savings can be achieved by using NFV, since virtual network appliances can run in a cloud-based environment, with hardware residing in totally geographically dispersed points of interest. NFV derives from Server Virtualization, a technique that enables users to share the resources and functionalities of a server concurrently through Virtual Machines (VMs) emulating a standard server. In a similar way, NFV migrates the standard network functions to a virtual or cloud based architecture, leading to high flexibility. NFV, IoT (Internet of Things), SDN (Software Defined Networking), SON (Self-Organized/Optimized Networks) and 5G are predicted to shape the future of telecommunication networks as many analysts, research centers and vendors claim. The role of NFV in this path is possibly the most important, since the vast traffic volume expected to be created through IoT connected devices and 5G networks cannot be efficiently and cost effectively supported with existing infrastructure deployments. Moreover, NFV is tightly coupled with SDN, with the latter having less stringent requirements as far as real time processing is concerned. Their combination can lead to flexibility with the development of non-proprietary communication protocols and the automation of network functions. The main difference between the two is that NFV aims at the decoupling of software and hardware, while SDN aims at the separation of packets and interfaces from the network control plane. In the first chapter of this thesis we present the background for network function virtualization, the key benefits that it can bring to the operators and telecommunication service providers as well as the correlation with software defined networking (SDN). Several standardization and architecture design approaches are described as well. In the second chapter we present the evolution from 2G (GSM) to 3G (UMTS) and 4G (LTE) and describe the main functions of the core and LTE radio network, since its specific elements are very good candidates for virtualization. We focus on LTE networks, since virtual EPCs (LTE core network) as well as virtual eNodeBs (Radio Access Part) are already deployed and operators and vendors understand that there shall be a clear benefit from virtualizing these elements due to their flat IP based architecture, simplicity of interfaces and CAPEX and OPEX savings that virtualization can provide. The third chapter describes the key drivers behind the virtualization of EPC as well as the basic architectures proposed by research and industry while the fourth chapter refers to the idea of virtualization in the radio access part of mobile networks deployed through C-RAN. We present the main reasons that led to the need for C-RAN deployments including a CAPEX and OPEX analysis and the basic deployment architectures up to this point in time.
Network Function Virtualization (NFV)
Mobile networks
Long-Term Evolution networks (LTE)
Capital Expenditure (CAPEX)
Operating Expense (OPEX)