Abstract:
Massive multiple-input multiple-output (MIMO) is an emerging technology, which is an enabler for future broadband wireless networks that support high speed connection of densely populated areas. Application of massive MIMO at the macrocell base stations in heterogeneous networks (HetNets) offers an increase in throughput without increasing the bandwidth, but with reduced power consumption. This research investigated the optimisation problem of signal-to-interference-plus-noise ratio (SINR) balancing for macrocell users in a typical HetNet scenario with massive MIMO at the base station. The aim was to present an efficient beamforming solution that would enhance inter-tier interference mitigation in heterogeneous networks. The system model considered the case of perfect channel state information (CSI) acquisition at the transmitter, as well as the case of imperfect CSI at the transmitter. A fast converging beamforming solution, which is applicable to both channel models, is presented. The proposed beamforming solution method applies the matrix stuffing technique and the alternative direction method of multipliers, in a two-stage fashion, to give a modestly accurate and efficient solution. In the first stage, the original optimisation problem is transformed into standard second-order conic program (SOCP) form using the Smith form reformulation and applying the matrix stuffing technique for fast transformation. The second stage uses the alternative direction method of multipliers to solve the SOCP-based optimisation problem. Simulations to evaluate the SINR performance of the proposed solution method were carried out with supporting software-based simulations using relevant MATLAB toolboxes. The simulation results of a typical single cell in a HetNet show that the proposed solution gives performance with modest accuracy, while converging in an efficient manner, compared to optimal solutions achieved by state-of-the-art modelling languages and interior-point solvers. This is particularly for cases when the number of antennas at the base station increases to large values, for both models of perfect CSI and imperfect CSI. This makes the solution method attractive for practical implementation in heterogeneous networks with large scale antenna arrays at the macrocell base station.
