Recently, with the rapid development of wireless communication technology, international standard-
ization work is on the way for the upcoming fifth generation(5G) mobile communication. Fifth generation
mobile communications should provide up to 20 times peak data rate, 100 times area traffic capacity,
and 10 times connection density than existing fourth generation(4G) mobile communications. In order
to satisfy these 5G core performance indicators, the ultra dense network(UDN) architecture should be
considered. In the case of UDN, area traffic capacity can be increased by reducing the cell size and closely installing the base stations, reliability can be also increased through efficient interference cancellation, and high transmission capacity can be provided with massive MIMO technology. However, connecting all the small-cell base stations to the backhaul network with conventional wired optical cable results in prohibitive cost, and it is difficult to support the high-speed network in a place where optical cable is difficult to install. In order to solve this problem, many researches have been actively studied to connect wirelessly a backhaul network of small-cell base stations.
The ultra wide area wireless backhaul system features multiple antenna structure of wireless backhaul-
hub and access-points, massive information transmission, wide cell coverage, and massive connectivity.
Utilizing this multiple antenna structure, the spatial multiplexing gain, the diversity gain, and the beam gain can be obtained. Through this gain, large capacity transmission, efficient interference cancellation, pathloss overcoming and reliable transmission can be achieved. However, such a large array antenna system causes high channel estimation complexity, feedback overhead, and system complexity, which makes actual implementation difficult.
In order to solve these problems, a hybrid beam forming technique which reduces the hardware
complexity by reducing the number of radio frequency(RF) chains has been studied, and some researches
are underway to reduce the amount of channel information feedback by utilizing the covariance channel,
which is the second-order statistics. However, in the case of these researches, there is a limitation in
providing a large-capacity transmission to each multiple user due to considering a system in which SU-
MIMO with multiple streams or MU-MIMO with a single stream for each user. In addition, in order
to apply the above studies to the ultra wide area wireless backhaul network, it is necessary to apply
the beamforming technology to compensate the pathloss due to the wide area support, and precoding
technology should be considered to efficiently remove inter-beam interference.
In this thesis, we proposed a transmit and receive beamforming scheme that can reduce the complex-
ity of the system while reflecting the characteristics of the ultra wide area backhaul network as mentioned above. To do this, the wireless backhaul-hub performs three-stage beamforming, and the access-point performs one-stage beamforming. In the first-stage beamforming of the wireless backhaul-hub, it is
designed to radiate a fixed analog beams to the supported area without reflecting the instantaneous
channel information. These analog beams was intended to overcome the pathloss. In the second-stage
beamforming, we designed a digital beamforming block that eliminates inter-access-point interference
by leveraging the effective channel matrix whose size was reduced through the backhaul-hub first-stage
beamforming and the access-point beamforming. In the last third-stage beamforming, a MIMO process-
ing block is designed to enable multiple stream transmission between the wireless backhaul-hub and each
access-point. In the case of the access-point, eigen-beamforming is performed in the dominant eigenvalue
direction of the effective channel. For each case of the uplink and downlink of the proposed system, the
average spectral efficiency and the bit error rate performance were analyzed by tracking the eigenvalue
distribution of the effective channel through each beamforming block, and the analytical results were
compared with the simulated results to verify the accuracy.