Back to search

National and Kapodistrian University of Athens

QoS-aware AI-empowered resource orchestration towards distributed, privacy-aware and sustainable wireless communications

Abstract

dc:description

The rapid evolution of mobile communication systems is steering the world towards a new era, supporting innovative and challenging use cases and applications such as fully autonomous vehicles, pervasive connectivity and digital twins. 5G Advanced, followed by further evolution to 6G, is expected to advance the capabilities of existing potential technologies, offering increased speed, highly reliable low-latency communications, massive connectivity, and intelligent system coordination. This shift will also support the development of more intelligent environments and infrastructures, where connectivity, automation, energy efficiency, and resource management are of paramount importance. As a result, it is becoming increasingly evident that next-generation networks must integrate a number of key enabling technologies including innovative energy-efficient architectures, advanced AI-driven network management, integrated sensing, and communication capabilities. The incorporation of AI-enabled edge computing and the expansion of distributed networks as a critical enabler toward supporting real-time processing requirements, especially for latency-sensitive applications, is anticipated. These emerging use case scenarios, applications and technologies will inevitably give rise to a number of new challenges, including those related to network reliability, data privacy, and energy efficiency. The proliferation of devices and the expanding scope of connected applications bring unique requirements for the network leading to a consistent adherence to QoS requirements, even under the most demanding network conditions. The increasing tendency in device connectivity will also lead to massive data generation and transfer throughout the network, especially in beyond 5G networks where AI will act as a key enabler. Collecting and analyzing vast amounts of heterogeneous data will inevitably raise privacy concerns. Additionally, as the demand for data and connectivity increases, it becomes apparent that so does the energy consumption of next generation cellular networks. Towards this direction, the current thesis introduces AURORA, a QoS-aware AI-empowered resource orchestrator designed to address the emerging challenges. By leveraging cutting edge technologies in the field of AI and predictive analytics, AURORA directly contributes to data privacy preservation, network reliability, energy efficient resource allocation and proactive service management. This thesis presents six key contributions to support the new requirements foreseen in next generation mobile networks. The first key contribution of the current thesis, is a privacy-aware distributed AI framework for Quality of Service (QoS) prediction. This framework represents a novel approach to predicting QoS for end users in a timely and accurate manner while addressing critical challenges such as data privacy, signaling overhead, and computational efficiency. To realize this, this framework shifts the prediction functionality from the core network to the edge, enabling edge network entities and end users to collaborate in the prediction process. To ensure privacy preservation, the framework incorporates techniques to encode any raw input data into highly complex, compressed, and irreversible latent representations before any transmission. The second significant contribution of this thesis is a comprehensive framework for service placement, addressing the critical challenge of efficiently allocating Virtual Network Functions (VNFs) and their associated traffic paths across the underlying network infrastructure. The proposed framework integrates intelligent orchestration that ensures alignment with resource constraints and optimal routing of traffic between VNFs, while also minimizing end-to-end delays. This contribution enables the delivery of high-quality services in next-generation networks, where resource optimization and performance guarantees are crucial. The third contribution of this thesis addresses the critical challenge of service scaling by providing a mechanism to dynamically and proactively adjust resource allocation in response to future traffic loads and service requirements. This framework monitors network conditions and proactively anticipates changes in resource requirements, enabling seamless service autoscaling. By incorporating real-time traffic analysis and predictive capabilities, the framework ensures that sufficient resources are allocated to maintain service quality. The fourth contribution introduces an intelligent energy saving framework that optimized the energy efficiency of distributed AI processes in wireless communication networks, addressing challenges such as high energy consumption, resource utilization inefficiencies, and environmental impact. The framework orchestrates computational and communication resources, emphasizing on device heterogeneity, data distribution, and complexity of the examined distributed AI framework. It also incorporates a simulated distributed AI environment, allowing for energy-efficient solution exploration without reducing the high computational and energy related costs that a real-world execution requires. The fifth contribution is SIM+, a holistic and implementation-agnostic Information Model (IM) designed to support AI-driven optimizations in forthcoming Beyond 5G network systems. SIM+ addresses the emerging challenges by standardizing data formats, enabling interoperable communication between network entities, and simplifying infrastructure complexities through a hierarchical abstraction. The flexibility of SIM+ is demonstrated through its application to two prominent AI-driven solutions. The last contribution of this thesis is a potential approach for deploying the AURORA scheme within existing cellular architectures. Given the ongoing evolution of architectural design for both 5G Advanced and 6G networks, and the lack of a standardized architecture, this contribution uses the 5G system (5GS) architecture as a reference point. It proposes a conceptual integration referencing the relevant network functions at the core, edge and RAN and how each one can contribute to the AURORA scheme.

Degree

thesis:*
Grantor dc:publisher
National and Kapodistrian University of Athens
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Magoula, Vasileia
  • Μαγουλά, Βασιλεία

Subjects

dc:subject × 22

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/60654

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Magoula, Vasileia; Μαγουλά, Βασιλεία. QoS-aware AI-empowered resource orchestration towards distributed, privacy-aware and sustainable wireless communications. National and Kapodistrian University of Athens, 2025. http://hdl.handle.net/10442/hedi/60654