Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Master of Engineering Science


Electrical and Computer Engineering


Shami, Abdallah


The evolution of fifth-generation (5G) and Beyond mobile technologies is spurred by the rapid demands and high-end requirements of next-generation mobile networks. It is imperative that advanced intelligence and machine learning techniques address these dynamic requirements by supporting network operations in terms of maintenance, servicing, and performance. The Third Generation Partnership Project (3GPP) has outlined a Network Data Analytics Function (NWDAF) for 5G Core (5GC) networks that should provide predictive network maintenance and improve network performance in these dynamic networks, and which must leverage the capabilities of artificial intelligence, machine learning, and advanced data analytics methods to satisfy its specification requirements. The work presented in this thesis surveys the current trends and future outlooks for 5G Core networks, in addition to presenting the capabilities of an implemented NWDAF, in emulated 5G environments, towards addressing a scaling optimization problem for Network Functions (NFs) in the 5G Control Plane. The insights from the NWDAF and its support in analytical and optimization problems justify its use as more than a network monitoring and data aggregation tool, but as an intelligence engine that will drive 5G and Beyond networks to satisfy user demand and improve consumer experience altogether.

Summary for Lay Audience

Modern networks continue to brace for an increasing number of mobile devices such as smartphones, cars, and smart home appliances/accessories connected in the Internet-of-Things (IoT), the tightly woven network of wireless devices and embedded systems inside the home and everywhere else. Mobile network providers and Internet Service Providers (ISPs) are faced with the difficult mission of preparing future networks to accommodate user demand in addition to providing the utmost experience and achieving their Quality-of-Service (QoS) metrics. Enabling technologies, such as Network Function Virtualization (NFV) and Software-Defined Networking (SDN), allow these networks to efficiently minimize costs of operations and maintenance, while at the same time, improving network performance; however, service providers must also consider the issues of their integration, including placement within the network, reliability of provided services, and the guarantee for high-importance applications’ needs to be constantly met. Modern and future networks are exploring new inter-network functionalities that are focused on data analytics and new services tailored towards advanced operations and maintenance. This is what is referred to as intelligent networking: it elucidates the ability of the network to recognize events of congested network traffic or points of failure and formulates decisions for the network (predictive maintenance) to accommodate these new requirements, through instantiation of new network function instances for example. These intelligent networking techniques leverage the use of machine learning, artificial intelligence, and advanced data analytics techniques to aid networks in their operations for the purpose of improving overall performance as well as user experience. The methodology of intelligent networking, as mentioned, can be categorized into a subfield of mathematics known as operations research. The work presented here demonstrates the use of network optimization models and statistical decision analysis to improve network capabilities through simulation: specifically, a Network Function (NF) scaling problem is addressed, involving how many instances of the NF are required to best serve the network and/or end users. The goal of this research is to justify the use of advanced intelligence as working engines in future networks for the purpose of improving performance and satisfying customer demands as required.

Available for download on Monday, July 15, 2024