National and Kapodistrian University of Athens
Design of cloud-native communications networks in mobile environments
Abstract
dc:descriptionThe advent of the 5th generation (5G) and beyond mobile networks will play a key role in addressing fundamental challenges related to a sustainable societal transformation and economic growth. These networks require a design that will support new service capabilities, enabling connectivity for a large number of devices, ubiquitous access with increased levels of mobility in heterogeneous environments and mission critical services with stringent performance requirements that need to be met in a cost-effective and energy-efficient way. However, to support these demanding and diverse services a paradigm shift is necessary that will enable migration from closed purposely developed infrastructures into open and elastic ecosystems. This can be achieved adopting flexible architectural models, and facilitating convergence of compute and network technologies. To this end, this thesis aims to exploit various architectural enhancements and technologies adopted from 5G/Beyond 5G (B5G) networks to design and deploy systems that can efficiently provide services in environments with varying levels of mobility. More specifically, it supports the idea of a universal 5G platform hosted in a cloud infrastructure that comprises interconnected compute, network and storage components. This platform is enhanced with suitable mathematical models, protocols and algorithms that are adapted and implemented with the aim to optimize the overall system performance. The thesis includes three distinct contributions that are implemented and experimentally validated through the use of an open-source toolset. The first contribution examines the architectural options provided by 5G/B5G networks and proposes efficient resource allocation strategies to enable optimal service delivery while addressing user mobility requirements. It introduces optimal resource management across various network segments, including the Radio Access Network (RAN), Core Network (CN), and Data Network (DN). These resources are virtualized and disaggregated to enhance flexibility in Network Function (NF) placement. To accommodate user mobility, this approach incorporates the feature of live Virtual Machine (VM) migration, which allows VMs hosting critical 5G functionalities to relocate dynamically as users move, ensuring uninterrupted service delivery. Furthermore, a multistage optimization framework is developed to achieve optimal allocation of both network and computational resources while minimizing operational costs. The analysis is supported by real-world mobility statistics and lab-based profiling measurements obtained from a 5G cloud testbed. The findings highlight trade-offs between latency and infrastructure costs, identifying optimal operational points for various scenarios. The second contribution focuses on the development of a multi-access connectivity system that integrates both 3GPP and non-3GPP access network technologies to support a diverse range of mobile User Equipment (UE). In this system, UEs with varying mobility patterns connect to the network, requesting the establishment of end-to-end (E2E) slices with the DN. When UEs connect via the 3GPP access network, their E2E slices are formed along paths that interconnect gNodeBs (gNBs), User Plane Functions (UPFs), and DN nodes. Conversely, for UEs opting for non-3GPP networks, connectivity to the DN is achieved through the Non-3GPP Interworking Function (N3IWF), which ensures secure communication and seamless integration into the network architecture. To address the optimal network selection problem, a resource allocation policy is proposed that aims to jointly minimize blocking rates for slow-moving UEs and dropping rates for fast-moving UEs. This approach is analytically evaluated using a two-dimensional Markov chain model. Additionally, the system's performance is validated through the cloud-based 5G testbed, providing key insights into E2E performance and resource consumption. Finally, in the third contribution a Management and Orchestration (MANO) framework is developed which is specifically designed for the orchestration operations of 5G networks. This framework emphasizes LifeCycle Management (LCM) of 5G components to establish an operational environment with minimal human intervention or manual configuration, aligned with the Zero-touch network (ZTN) Service Management (ZSM) paradigm. At the core of this framework an Artificial Intelligence (AI)/Machine Learning (ML) block with monitoring capabilities is present, performing decision-making across various layers of the infrastructure. This platform optimizes the allocation and orchestration of both networking and edge/cloud computing virtual resources, enhancing the overall efficiency and adaptability of the system. The proposed framework is validated through two use-cases. The first demonstrates the dynamic deployment of network slices on a softwarized, multi-operator, lab-based 5G platform. The second showcases a proactive User Plane Function (UPF) provisioning mechanism, which enables the system to predict future compute and network demands of a slice and dynamically adjust resource allocation to meet those demands effectively.
Degree
thesis:*- Grantor dc:publisher
- National and Kapodistrian University of Athens
- Year dc:date
- 2024
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Manolopoulos, Alexandros-Ioannis
- Μανωλόπουλος, Αλέξανδρος-Ιωάννης
Subjects
dc:subject × 17- Δίκτυα 5ης γενιάς
- Διαχείριση κινητικότητας
- Εικονικοποίηση δικτυακών λειτουργιών (NFV)
- Δίκτυα μηδενικής παρέμβασης
- Υλοποίηση δικτύων σε υπολογιστικά νέφη
- 5G
- B5G
- Mobility management
- Network function virtualisation (NFV)
- Zero touch network and service management (ZSM)
- Cloud mobile networks
- Επιστήμη Ηλεκτρονικών Υπολογιστών και Πληροφορική
- Φυσικές Επιστήμες
- Δίκτυα ηλεκτρονικών υπολογιστών και Επικοινωνίες
- Computer and Information Sciences
- Natural Sciences
- Computer Networks and Communications
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/58637