Abstract
The multiuser multiple-input multiple-output (MU-MIMO) antenna system is a key technology of Long Term Evolution-Advanced (LTE-A), which
can achieve much higher data rates with limited bandwidth. However, there is a trade-off between the system capacity gain achieved by the MUMIMO
system and the quality of service (QoS) requirements, e.g. latency, throughput guarantee and delay variation. In this thesis, our main
objective is to develop a QoS-enabled MU-MIMO system that can support larger numbers of users with QoS guarantees. In particular, we focus on
designing QoS-aware scheduling and resource allocation algorithms for MU-MIMO which can be deployed with practical restrictions in an LTE-A
network.
The first contribution of this thesis is the design of QoS-aware user scheduling for a downlink MU-MIMO system to provide a good trade-off among
the system throughput, fairness and delay restrictions. We design an efficient priority metric which exploits the delay information and the
encoding decoding rate of the video streaming. Based on this priority metric, we propose a complexity-reduced user scheduling algorithm.
The second contribution of this thesis is a comprehensive solution for joint user grouping and resource allocation in a downlink MU-MIMO system
with a frequency-selective channel model. We provide useful insights into the complexity of exploring each sub channel and then propose lowcomplexity
algorithms that efficiently bundle the sub channels over which potential user groups are likely to experience similar channel conditions.
We provide a feedback scheme that scales down the feedback overhead without sacrificing much in the QoS.
The third contribution of this thesis is the design of low-complexity scheduling and resource allocation of heterogeneous users for an uplink MUMIMO
system. The goal is to maximize the system throughput while maintaining delay bounds for delay-sensitive traffic. This optimization problem
turns out to be a three-dimensional assignment problem which in general can be solved by the exhaustive search method. We propose suboptimal
algorithms in which the key ideas are to reduce the search space and iteratively minimize the rate loss. We design an efficient method which
dynamically derives a minimum required data rate to satisfy the defined delay bound.