Towards a Service-aware E2E Network Slice Orchestration and Resource Provisioning across Multiple Domains

Abstract

The 5th generation mobile network will rely fundamentally on network slicing to provide the required support to enable different vertical industry use cases, such as Ultra Reliable Low Latency Communication (uRLLC), Enhanced/extreme Mobile BroadBand (e/xMBB) and Massive Machine Type Communication (mMTC). In order to deliver these variant 5G use cases to end users from the same physical infrastructure, slice providers have to ensure the customization and optimization of End-to-end (E2E) mobile network slices to truly reflect the type of services that will be running on them. This dissertation is a compilation of research outcomes from our investigations into the practical implementations of various infrastructures and theoretical architectural designs. We also propose a set of algorithms to support the orchestration of service-aware network slices to enable different 5G and beyond use cases.

Network function virtualization and software-defined networking are fundamental enabling principles in network slice orchestration. The orchestration of network slices that cuts across the different technological domains of the mobile network (i.e., RAN, Transport, and Core network) as well as the administrative domains (i.e., different cloud providers, MNO clouds and other private/public clouds) is called E2E network slicing. Orchestrating E2E network slices to deliver different 5G services involves a number of requirements and constraints at different technological domains of the mobile network. Fulfilling these divergent but distinctive network requirements for each slice type while provisioning virtual resources from multiple administrative domains will require a holistic framework and intricate architecture capable of considering all of the possible slicing ramifications and the necessary system support. Traffic isolation, dynamic resource provisioning, and dynamic re/configuration are among the most notable challenges of network slicing.

The aim of the studies presented in this dissertation is to make advances towards overcoming the specific challenges of network slicing. These challenges have been approached from the perspectives of a sound theoretical analysis, the implementation of a practical framework, the design and development of novel architecture, and the development of efficient schemes. We started by conducting a comprehensive survey to provide a holistic understanding of the concept of network slicing, its design principles, and enabling technologies. Then, we implemented a framework for the E2E slicing of the mobile network and 5G services. We then proposed a novel architecture for facilitating the orchestration of network slices from multiple administrative domains. Finally, we developed a scheme with an enabling architecture for the dynamic provisioning of virtual resources for slice orchestration systems.