Containerised NFV orchestration and management

Network Function Virtualization (NFV) has been gaining traction as a way to manage computer networks, and to enable fast and flexible service deployment on top of physical connectivity. Lightweight virtualization approaches such as, e.g., software containers, are also being increasingly explored as a platform for efficient orchestration and management of virtualised Network Functions at scale.

Current networks operations and management (O&M) solutions assume software containers are deployed within data centres with single geographic location, high speed communications and high performance infrastructure. However, many network benefits can be obtained if container network functions can be distributed around the network edge.

This PhD research will address the challenge of how to devise network-wide platforms that can manage and interconnect Network Functions for many customers at scale and over the Internet. We have been using platforms like Openstack as well as our own Glasgow Network Function (GNF) orchestrator, and we will further investigate how such platforms can be designed to work at scale (e.g., billions of small IoT devices) and over heterogeneous environments where network resource utilisation becomes as significant as CPU and memory resource use.

There is a rich bibliography regarding the mathematical interpretation of orchestrating services and managing resources. An adaptation of this work to the needs of modern networking systems can lead to significant results. The evaluation of this adaptation will be done through simulation and also through deployment on actual testbed systems.

In NFV, the software is decoupled from hardware allowing the migration of services according to resource availability or specific resource constraints. This enables an efficient management of resources and reduces the Operational Expenditure. This research will extend the work of Open MANO and working on the evolution of ICT networks and distributed systems towards the most energy efficient way of operation. The means to achieve that is by making use of applied probabilities and statistics. By developing an intelligent way of utilizing network resources and systems, we aim to contribute towards the Connected Nation prosperity outcome.

Network functions can be used for accurate monitoring of network parameters through analysing real-time traffic or hardware characteristics like utilization, temperature, etc. By using statistical models, malicious traffic or hardware changes can be detected, and a potential attack can be promptly identified. This direction of research helps establish a safe cyber society and is well aligned with EPSRC's Resilient Nation prosperity outcome.

In addition, virtualised computing infrastructure operating at the Edge of the network are being investigated for the development of smart-home and assisted leaving technologies such as health and safety monitoring, environmental and personal condition monitoring and alerting, etc. Therefore, the infrastructure developed within this project will also be contributing towards the Healthy Nation prosperity outcome.

Overall, the goal of this project is to advance networking communications by making intelligence an integral part of NFV management and orchestration. The research will be conducted in close collaboration with BT who sponsor this PhD through an iCASE award, and want to explore lightweight virtulisation technologies for the efficient management of customer premises equipment.