@PhdThesis{duepublico_mods_00044636,
  author = 	{Gao, Yuan},
  title = 	{Low RF-complexity massive MIMO systems: antenna selection and hybrid analog-digital beamforming},
  year = 	{2017},
  month = 	{Oct},
  day = 	{11},
  keywords = 	{Massive MIMO; antenna selection; hybrid analog-digital beamforming},
  abstract = 	{Wireless data traffic has been increased dramatically in the last decades, and will continue to increase in the future. As a consequence, the infrastructure of wireless communication systems needs to advance on the data capacity. Massive Multiple-Input Multiple-Output (MIMO) is a promising candidate technology to meet the demand. By scaling up the conventional MIMO by orders of magnitude number of {\backslash}emph{\{}active{\}} antennas, a massive MIMO system can harvest considerable channel degrees of freedom to increase the spectral efficiency. However, increasing the number of {\backslash}emph{\{}active{\}} antennas needs to increase both the numbers of Radio Frequency (RF) transceivers and antenna elements {\backslash}emph{\{}at the same rate{\}}, which will increase the RF complexity and cost dramatically. It is known that the complexity and cost of antenna elements are usually much lower than that of RF transceivers, which motivates us to scale up MIMO by a lower increasing rate of the number of RF transceivers than that of antenna elements, resulting in so-called low RF-complexity massive MIMO systems. In this thesis, we study two types of low RF-complexity massive MIMO systems, i.e., massive MIMO antenna selection systems and massive MIMO hybrid analog-digital beamforming systems. Both systems use specific RF networks to bridge a massive number of antennas and a small number of RF transceivers, leading to signal dimension reduction from antennas to RF transceivers. The RF network used in antenna selection is referred to as RF switching network; while the RF network used in hybrid beamforming is referred to as Phase Shifting Network (PSN). Both RF networks have two types of architectures, i.e., full-array architecture and sub-array architecture. The latter has lower insertion loss, lower complexity and better scalability than the former, but at the price of performance degradation caused by connection constraint, which will be studied for both low RF-complexity systems in this thesis. In addition, a low RF-complexity PSN for the hybrid analog-digital beamforming system needs also to be studied to replace the conventional high-complexity-and-cost phase-shifter-based PSN.

In the antenna selection system, the upper bounds on the channel capacity using asymptotic theory on order statistics are derived at the large-scale limit. The optimal antenna selection algorithms are also developed, which are based on Branch And Bound (BAB) search algorithm. Through the theoretical and algorithm studies, it is found that the sub-array antenna selection has close performance to the full-array antenna selection. In the hybrid beamforming system, we propose to use Rotman lens as PSN, which is of lower complexity and cost than the conventional phase-shifter-based PSN. Two beam selection algorithms, i.e., sub-optimal greedy search and optimal BAB search, are also proposed. In addition, the Rotman lenses are designed, fabricated and measured. The measurement results together with the beam selection algorithms are used to perform Monte Carlo simulation. Simulation results show that the proposed Rotman-lens-based system with the sub-array architecture suffers noticeable performance degradation compared to the system with the full-array architecture when ideal Rotman lenses are used. But when practical non-ideal Rotman lens are used, the former outperforms the latter when the number of antennas is large enough. Most interestingly, with non-ideal hardware, the sub-array Rotman-lens-based system has close performance to the sub-array phase-shifter-based system, and also exhibits a wideband capability. To prove the advantage of the low RF-complexity massive MIMO, two testbeds are built up for the antenna selection and hybrid beamforming systems, respectively. The measurement results show the low RF-complexity massive MIMO systems have superior performance over the small-scale MIMO systems under the condition of the same number of RF transceivers. 

The results in this thesis show that the low RF-complexity massive MIMO systems proposed in this thesis are feasible in technology and promising in performance, validating its potential usage for the future 5G wireless communication systems.},
  url = 	{https://duepublico2.uni-due.de/receive/duepublico_mods_00044636},
  file = 	{:https://duepublico2.uni-due.de/servlets/MCRFileNodeServlet/duepublico_derivate_00044130/Diss.Gao.pdf:PDF},
  language = 	{en}
}