Publicação
Cell-free networks with hybrid equalization: radio stripes vs. centralized and distributed architectures
| Resumo: | Cell-free networks are a paradigm shift from the current cellular network, proposing a novel solution to inter-cell interference through the usage of small access points physically spread throughout a specific area, all connected to a processing unit, which allows the co-processing of user signals. However, due to cell-free networks being a recent development, there are many design questions to be answered, with the focus of this work being the architecture which determines how signal processing occurs during uplink transmission. This can be achieved through three distinct architectures, the centralized one has every access point relay their signals to the processing unit, while the distributed one has each access point process their received signals and then combine them in the processing unit and finally the radio stripe system where each access point is connected to another, allowing for serial processing of the user signals. Cell-free also enables the use of millimeter wave technology to provide high bit rates, and massive MIMO with the large number of access points. These are expected in any 6G network, but can not be simultaneously deployed in a realistic scenario without hybrid beamforming. The three architectures were modeled and then simulated to obtain their relative performance, with distributed falling behind the other two due to the diminished co-processing, while the other architectures are on par in terms of performance, highlighting the radio stripe system due to its other benefits for realistic deployments and cost reduction. Hybrid beamforming was also modeled, showing that it does incur a performance loss when comparing systems with analog equalizer stages to those without, but this gap can be greatly diminished when the equalizer is developed for a particular system such as was done to radio stripes. For these reasons, the radio stripe system was chosen to be simulated in various scenarios to determine the performance effects of a few relevant parameters on the network. Access point and receiver antenna increases displayed the greatest gains at the cost of added computational complexity, while the number of user terminal and radio frequency chains were shown to be related. These last two can lead to significant performance implications if the system is not scaled properly, requiring careful consideration to have an adequate amount of chains for a given number of users. |
|---|---|
| Autores principais: | Leite, Miguel Tavares |
| Assunto: | Cell free Massive MIMO mmWave Hybrid beamforming Radio stripe |
| Ano: | 2024 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade de Aveiro |
| Idioma: | inglês |
| Origem: | RIA - Repositório Institucional da Universidade de Aveiro |
| Resumo: | Cell-free networks are a paradigm shift from the current cellular network, proposing a novel solution to inter-cell interference through the usage of small access points physically spread throughout a specific area, all connected to a processing unit, which allows the co-processing of user signals. However, due to cell-free networks being a recent development, there are many design questions to be answered, with the focus of this work being the architecture which determines how signal processing occurs during uplink transmission. This can be achieved through three distinct architectures, the centralized one has every access point relay their signals to the processing unit, while the distributed one has each access point process their received signals and then combine them in the processing unit and finally the radio stripe system where each access point is connected to another, allowing for serial processing of the user signals. Cell-free also enables the use of millimeter wave technology to provide high bit rates, and massive MIMO with the large number of access points. These are expected in any 6G network, but can not be simultaneously deployed in a realistic scenario without hybrid beamforming. The three architectures were modeled and then simulated to obtain their relative performance, with distributed falling behind the other two due to the diminished co-processing, while the other architectures are on par in terms of performance, highlighting the radio stripe system due to its other benefits for realistic deployments and cost reduction. Hybrid beamforming was also modeled, showing that it does incur a performance loss when comparing systems with analog equalizer stages to those without, but this gap can be greatly diminished when the equalizer is developed for a particular system such as was done to radio stripes. For these reasons, the radio stripe system was chosen to be simulated in various scenarios to determine the performance effects of a few relevant parameters on the network. Access point and receiver antenna increases displayed the greatest gains at the cost of added computational complexity, while the number of user terminal and radio frequency chains were shown to be related. These last two can lead to significant performance implications if the system is not scaled properly, requiring careful consideration to have an adequate amount of chains for a given number of users. |
|---|