Proposal and evaluation of connectivity solutions for beyond 5G radio access networks
(English) In the context of the fifth generation (5G) of mobile communications, the number of connected devices is expected to increase substantially compared to previous systems (e.g., LTE). Similarly, more stringent user requirements in terms of quality of service (QoS) are anticipated, beyond hig...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2024 |
| País: | España |
| Institución: | CBUC, CESCA |
| Repositorio: | TDR. Tesis Doctorales en Red |
| OAI Identifier: | oai:www.tdx.cat:10803/693737 |
| Acceso en línea: | http://hdl.handle.net/10803/693737 https://dx.doi.org/10.5821/dissertation-2117-425479 |
| Access Level: | acceso abierto |
| Palabra clave: | Àrees temàtiques de la UPC::Enginyeria de la telecomunicació 621.3 |
| Sumario: | (English) In the context of the fifth generation (5G) of mobile communications, the number of connected devices is expected to increase substantially compared to previous systems (e.g., LTE). Similarly, more stringent user requirements in terms of quality of service (QoS) are anticipated, beyond high data rates (e.g., latency, reliability, etc.). Providing these services adequately represents a challenge for mobile network operators in terms of network deployment and operation. Therefore, adopting strategies to ensure adequate connectivity conditions is critical for mobile network operators to successfully provide 5G and future (6G) services. In recent years, artificial intelligence (AI) has revolutionized mobile network design, deployment and management processes, thus driving the development of innovative solutions, sustained also on other technological advances such as software-defined networking (SDN), network functions virtualization (NFV). In fact, the impact of these technological tools is not limited to 5G systems, but will also extend to future generations such as 6G. In this context, this thesis addresses the challenge of proposing, developing and evaluating solutions at the Radio Access Network (RAN) level with the objective of ensuring optimal connectivity conditions and thus satisfying the QoS requirements of network users. Various techniques are used to achieve this goal, with special emphasis on the use of AI techniques. Firstly, the thesis presents a model to optimize Multi-connectivity (MC) in heterogeneous networks. MC is a key technology for managing high traffic densities and meeting stringent QoS requirements. Performing effective multi-connectivity management is challenging due to various factors such as propagation conditions, interference, loads of various cells, QoS metrics, etc. To address these challenges, the thesis presents a novel algorithm designed to dynamically split User Equipment (UE) traffic between different RATs and cells. The algorithm aims to satisfy QoS requirements while minimizing radio resource consumption in order to minimize the possibility of congestion in the involved cells. The proposed solution is based on the Deep Q-Network (DQN) algorithm. Through a training phase, the model learns an optimal traffic splitting policy to be applied to each UE. The policy adapts to the current conditions of both the UE and the network. This adaptive approach improves network performance by increasing user throughput while mitigating the risk of having cell congestion. Deepening into the problem of ensuring adequate network coverage and as a second contribution, the thesis presents a methodology for coverage optimization in 5G systems. This methodology is based on two main tasks: detection and resolution of coverage holes. It is introduced a Machine Learning based model capable of detecting and characterizing coverage holes by analyzing real network traffic data. A coverage hole becomes significant for network performance when there is a persistent presence of users in its region, as this is reflected in a degradation of both user experience and overall network performance. As a solution to the coverage holes, the thesis considers the integration of relays in order to extend network coverage; the solution comprises both fixed and mobile relays (i.e., UEs acting as relays). The solution based on fixed relays focuses on a functionality to mitigate coverage holes by strategically placing relays in order to improve network availability. On the other hand, it is proposed a solution to extend coverage based on Relay UEs (RUEs). To achieve this effectively, a DRL algorithm is proposed for the intelligent activation or deactivation of the RUEs. Overall, the proposed coverage optimization methodology has demonstrated its feasibility, leading to improved network and user performance. |
|---|