Radio access network optimization with proactive resource management for 5G and beyond
(English) The Fifth-Generation (5G) of cellular networks significantly increases the performance and flexibility of the offered services to users and service providers. The strict network requirement of 5G use cases has been supported by integrating service-based architecture in the core network,fle...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis doctoral |
| Fecha de publicación: | 2023 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/400319 |
| Acceso en línea: | https://hdl.handle.net/2117/400319 https://dx.doi.org/10.5821/dissertation-2117-400319 |
| Access Level: | acceso abierto |
| Palabra clave: | Àrees temàtiques de la UPC::Enginyeria de la telecomunicació |
| Sumario: | (English) The Fifth-Generation (5G) of cellular networks significantly increases the performance and flexibility of the offered services to users and service providers. The strict network requirement of 5G use cases has been supported by integrating service-based architecture in the core network,flexible radio access network architecture, and implementing numerous wireless technologies. Researchers and Mobile Network Operators (MNOs) face vast challenges not only in the definition process but also in the deployment phase. The research community should define robust and dynamic radio network solutions to tackle the complexity and flexibility of 5G and beyond mobile network requirements. As mentioned, the radio access network architecture has been crucial in defining 5G systems. Especially the Cloud Radio Access Network (C-RAN) architecture has played a fundamental role as part of the new generation radio access network (NG-RAN) because it has the potential to support extremely dense radio network deployments while reducing costs because of the simplification of the radio units. Moreover, C-RAN enhances the network capacity by reducing the number of required resources because it centralizes the baseband functionalities in Baseband Units (BBU) pools or Central Units (CUs), sharing the same resource to manage multiple Remote Radio Heads (RRHs) or Radio Units (RUs). Moreover, Coordinated Multipoint (CoMP), enhanced Inter-Cell Interference Coordination (eICIC), and beamforming technologies could be easily implemented in the C-RAN structure, improving the 5G network performance. However, sustainability and energy efficiency are crucial to support future services and network requirements. Namely, It is fundamental to reduce energy consumption, resource usage, and emissions footprints to avoid excessive power consumption. The enormous increase in the number of devices, data amounts, and data rates implies an increase in the overall data traffic and required capacity, while energy reduction is not automatically guaranteed. On the other hand, the optimal management of the computational resources to satisfy current and future network requirements also becomes a challenge. This doctoral thesis aims to address some of the challenges above. Most of the published research works employ synthetic scenarios to validate the results. A realistic C-RAN platform has been implemented, opposing these approaches. The proposed architecture considers rural and urban zones, heterogeneous deployment with macro and small cells, time-variant traffic patterns, realistic user equipments with guaranteed bit rate and best effort services, Quality of Service (QoS) constraints, a 3D ray-tracing propagation model, multiple frequency bands, different split options, among other significant features. This platform becomes fundamental in the validation of the proposed algorithms. 5G and beyond radio network deployment will be ultra-dense. However, optimizing the costs and energy consumption is not automatically guaranteed. For this reason, this thesis also provides a non-linear data modeling and decision-making tool to maximize cost reduction versus coverage-QoS trade-off by optimizing the active RRHs needed according to traffic demands. Besides, it proposes a novel strategy for Dynamic Resource Management with Adaptive Computational capacity (DRM-AC) using Machine Learning (ML) techniques. Three ML algorithms have been considered in the final design after testing multiple approaches: support vector machine (SVM), time-delay neural network (TDNN), and long short-term memory (LSTM). DRM-AC reduces the average of unused resources by 96 %, but there is still QoS degradation when RCC is higher than the predicted computational capacity (PCC). To further improve, two new strategies are proposed and tested in a realistic scenario: DRM-AC with pre-filtering and DRM-AC with error shifting, reducing the average of unsatisfied resources by 98% and 99.9% compared to the DRM-AC. |
|---|