Zero-forcing coordinated beamforming in multi-cell MIMO systems: efficient algorithms and optimal feedback

Abstract

In this thesis, coordinated beamforming (CB) for multi-cell time-varying multiple-input multiple-output (MIMO)/multiple-input single-output (MISO) interference channels is investigated. In the future wireless communication systems like 3GPP LTE-Advanced standard, CB is one of the promising techniques to manage the inter-cell interference in the multi-cell environment, and thus the overall system capacity can be improved using CB techniques. Most previous works on CB is focused on the beam design problem in static channels. However, real wireless channels are time-varying, and this time-varying nature of channels has not been considered extensively for CB. Hence, in this thesis, the beam design problem and a limited feedback framework for zero-forcing CB(ZF-CB) in multi-cell time-varying MIMO interference channels is considered.

In the first part of the thesis, the problem of a beam tracking for ZF-CB in time-varying MIMO interference channels is considered. In time-varying channels, the beamforming matrices for ZF-CB should be redesigned when the channel information is changed. However, the calculation of ZF-CB matrices in every time steps needs the full subspace decomposition, and this leads huge computational complexity at each base station. To overcome this problem, a simple tracking algorithm for ZF-CB matrix is proposed. For this, the perturbation term for the null space of the fat matrices is first derived. Based on this perturbation results, the new beam tracking algorithm which simply updates beamforming matrices by additive manner is proposed. The computational complexity is analyzed, and it is shown that the proposed algorithm significantly reduces computational complexity. Also, the performance of the proposed algorithm is analyzed. For more generalized communication environments, multi-cell MIMO interfering broadcast channels (MIMO-IFBC) is considered, and the beam tracking algorithm for this case is also proposed. And, the computational complexit...