Mechanisms for integration of MEC and NFV for 5G networks in dynamic and heterogeneous scenarios

Thumbnail Image
Publication date
Defense date
Journal Title
Journal ISSN
Volume Title
Google Scholar
Research Projects
Organizational Units
Journal Issue
The 5G technology presents a significant leap into making the Information and Communication technology and integral part of the industries, and societies. Enhanced connectivity features unlock a range of different applications that provide unique user experience such as virtual and augmented reality, or mission-critical communications that improve the healthcare and environmental protection, digital twin for optimizing the production lines, etc. Besides the new radio technology, the virtualization technology is the major enabler of most of the exciting novel applications. Viritualization enables service providers to customize and shape the existing computing, networking and storage infrastructure to accommodate the requirements of the different range of customers often referred as vertical industries. The Network Function Virtualization (NFV) with the Software-defined Networking (SDN) are the key technologies that enable deployment of multiple isolated and customized networks on top of a single administrative domain infrastructure. The Multiaccess Edge technology revamps carrier’s infrastructure with application-oriented capabilities feeding applications with context information to elevate the user experience. Even though initially projected as a mobile operator technology, it is applicable to any service provider. This thesis departs from the point on how to integrate both, NFV and MEC, for different environments and scenarios. The MEC technology is not virtualized intrinsically hence the first part of the thesis explores the integration of MEC in NFV environment. Initially a MEC application and the utilization of radio context information is showcased through an Edge robotics scenario. Later the full integration of virtualized MEC components within an NFV infrastructure is elaborated through categorization, and proposed solution in tackling integration issues. Further, a tutorial is presented on how the exemplary Edge robotics would be deployed, terminated and managed in an MEC in NFV environment. The elaborated procedures present high compatibility and readiness for MEC in NFV future deployments. The findings are compared with existing works on the similar topic. The joint, or horizontal, integration of MEC in NFV is referred to a single administrative domain. The rest of the thesis is focusing on how administrative domains are able to fulfill vertical requirements by deploying end-to-end services across multiple domains. One of the thesis contribution is towards the definition, characterization and classification of federation - the process of deploying NFV services across multi-domain scenarios. Further the federation scenario is showcased in a static environment for a novel missioncritical eHealth application. All the federation functionalities are demonstrated in a realcase experimental emergency scenario for a patient suffering from a heart-attack. The assumption is that the federation occurs between two administrative domains enabling end-to-end AR/VR emergency services spread across two NFV based infrastructures. The obtained experimental results provide improvement in the future emergency events while leveraging on novel technologies such as AR/VR. The drawbacks are evaluated accordingly. The use of both MEC and NFV enables better user experience, customized networks and it is a big step towards automation, and reactive network life-cycle management. The following part of the thesis focuses on how to apply the federation concept in dynamic environments - where the conditions change rapidly, the resources are volatile and the relationships between administrative domains are established on-the-fly or unexpectedly broken. Blockchain as a Distributed Ledger Technology (DLT) is applied to facilitate and build trust in the brief negotiation process between mutually unknown administrative domains. A concrete step-by-step process is proposed which its application, in orchestration and life-cycle management (e.g., healing process), of emulated NFV service has been experimentally evaluated across multiple Blockchain platforms. Additionally, the Blockchain solution is applied in a small-scale Edge robotics experimental scenario. The Edge robotics service is a MEC-in-NFV based remote control application for mobile robots which leverage the DLT federation to extend the robot driving range by deploying radio network extension on top of an external domain infrastructure, without any interruption or downtime of the end-to-end Edge robotics service. In the last part of the thesis the focus is set on how service provides or telco operators may increase their profit margins by leveraging the federation process and using Machine Learning algorithms to generate a profitable decision of whether to federate a service or deploy the service over the constituent infrastructure. The application of Reinforcement learning algorithms such as Q-learning provides a promising near-optimal results. These are improved with the application of Deep Q-learning techniques through the use of real dynamic price fluctuations for service offerings.
5g networks, MEC, Multi-access edge computing, NFV, Network function virtualization, SDN, Software-defined networking, Federation, Machine-learning
Bibliographic citation