JultikaUniversity of Oulu repository

Abstract

Local 5G network are emerging as new form of 5G deployment targeted are service delivery for vertical specific purposes and other local users. As such, a well-defined network slicing architecture and implementation procedure is required for a local 5G network. A local 5G network also known as a 5G micro-operator network is targeted a network delivery for vertical-specific services. The aim of the micro-operator concept is to provide enough network flexibility and customization required by different vertical. Previous works on the micro-operator network have established different deployment scenarios that can exist, namely Closed, Open and Mixed Network. Thus, in order for any deployment of a micro-operator network to achieve the network flexibility, customization and privacy required by various vertical, it is essential to have a well-defined network slicing architecture and implementation procedure for local 5G networks. In this thesis, a sophisticated end-to-end network slicing architecture is proposed for different deployment scenarios of a local 5G micro-operator. The aim of the architecture is to address the unavailable description of network slicing for vertical specific network providers, leveraging multi-domain and multi tenancy. The proposed architecture incorporates a broad four-layer concept, leveraging a Multi-tenancy layer for different tenants and their end users, a descriptive Service layer, a multi-domain Slicing MANO layer and a Resource layer. A message sequence diagram is established based on the proposed architecture to describe the flow of information from when a tenant request a slice till the network slices are allocated as communication services to the various targeted user equipment. An actual implementation of network slicing is developed for specific layers of the proposed architecture. To do this, we used a softwarized network based on SDN/NFV, using OpenStack as a cloud infrastructure. On top of that, the network slicing implementation was done using the ETSI Open Source MANO. With these tools, different deployment scenarios’ implementations are achieved. Performance analysis are made based on metrics such as CPU utilization, memory utilization, rate of packet sent and packet received between different network service. These metrics are used to compare shared and non-shared slices within a single or multiple domain slice implementation, which were used as basis for classification of network slice instantiation in 5G micro-operator deployment scenarios. The results from the thesis successfully support the end-to-end network slicing architecture for various deployment scenarios of a local 5G micro-operator network, proposes a slice formation sequence from the end users to the micro-operator network for each deployment scenarios, implement different part of the architecture using different open source tools and measure the performance metrics of different deployment scenarios based on CPU or memory utilization.