Beyond Massive MIMO : Trade-offs and Opportunities with Large Multi-Antenna Systems
Sammanfattning: After the commercial emergence of 5G, the research community is already putting its focus on proposing innovative solutions to enable the upcoming 6G. One important lesson put forth by 5G research was that scaling up the conventional multiple-input-multiple-output (MIMO) technology by increasing the number of antennas could be extremely beneficial for effectively multiplexing data streams in the spatial domain. This idea was embodied in massive MIMO, which constitutes one of the major technical advancements included in 5G. Consequently, 6G research efforts have been largely directed towards studying ways to further scale up wireless systems, as can be seen in some of the proposed 6G enabling technologies like large intelligent surface (LIS), cell-free massive MIMO, or even reconfigurable intelligent surface (RIS). This thesis studies the possibilities offered by some of these technologies, as well as the trade-offs that may naturally arise when scaling up such wireless systems.An important part of this thesis deals with decentralized solutions for base station (BS) technologies including a large number of antennas. Already in the initial massive MIMO prototypes, the increased number of BS antennas led to scalability issues due to the high interconnection bandwidths required to send the received signals---as well as the channel state information (CSI)---to a central processing unit (CPU) in charge of the data processing. These issues can only be exacerbated if we consider novel system proposals like LIS, where the number of BS antennas may be increased by an order of magnitude with respect to massive MIMO, or cell-free massive MIMO, where the BS antennas may be located far from each other. We provide a number of decentralized schemes to process the received data while restricting the information that has to be shared with a CPU. We also provide a framework to study architectures with an arbitrary level of decentralization, showing that there exists a direct trade-off between the interconnection bandwidth to a CPU and the complexity of the decentralized processing required for fixed user rates.Another part of this thesis studies RIS-based solutions to enhance the multiplexing performance of wireless communication systems. RIS constitutes one of the most attractive 6G enabling technologies since it provides a cost- and energy-efficient solution to improve the wireless propagation links by generating favorable reflections. We extend the concept of RIS by considering reconfigurable surfaces (RSs) with different processing capabilities, and we show how these surfaces may be employed for achieving perfect spatial multiplexing at reduced processing complexity in general multi-antenna communication settings. We also show that these surfaces can exploit the available degrees of freedom---e.g., due to excess of BS antennas---to embed their own data into the enhanced channel.
KLICKA HÄR FÖR ATT SE AVHANDLINGEN I FULLTEXT. (PDF-format)