AbstractWe tend to assume that pervasive wireless connectivity exists everywhere, and with new applications such as Virtual Reality, Autonomous Driving and the Internet of Things, this with require a 1000-fold increase in capacity by 2020 based on cellular capacity in 2016. This increase now presents signiﬁcant challenges to wireless system designers. The majority of consumer wireless technologies currently operate using spectrum between 380MHz and 6GHz. Given the favourable propagation for cellular communications, wireless local area networking (Wi-Fi), terrestrial television broadcasting and radar, this spectral spot has become extremely crowded, and mechanisms to enhance the spectrum efﬁciency are thus needed. Massive multiple-input, multiple-output (Ma-MIMO) enables efﬁcient utilization of the limited radio resources providing signiﬁcant capacity improvement by multiplexing more simultaneous users within the same radio resource by means of spatial domain signal processing.
This thesis evaluates the practicality of using Ma-MIMO in real-world scenarios and identiﬁes solutions to operational deployments as they are uncovered. At the time this work commenced, only a few real-time evaluations had been conducted and the majority of the theoretical papers were not validated. The work in this thesis presents the per-cell spectral efﬁciency (SE) of 145.6 bits/s/Hz achieved with a 128-antenna Ma-MIMO testbed and introduces the challenges and the limitation factors that were revealed when delivering this level of spectrum efﬁciency enhancement. Here, the Ma-MIMO channel robustness is evaluated through various indoor and outdoor ﬁeld trials, and the improved channel robustness was then exploited to design a simple uplink closed loop power control scheme.
This thesis also introduces novel adaptive user grouping algorithms to further enhance the Ma-MIMO advantages for different applications. The algorithms presented address the interference caused by the multiple simultaneous user channel vectors in the same frequencytime domain as well as errors due to hardware impairments. This can be achieved by means of a novel error vector magnitude (EVM) prediction method proposed herein. Importantly, all the proposed algorithms and approaches in this thesis have been evaluated experimentally by means of a software-deﬁned radio Ma-MIMO testbed.
|Date of Award||19 Mar 2019|
|Supervisor||Angela Doufexi (Supervisor) & Mark A Beach (Supervisor)|