This paper proposes a new precoding approach that utilizes two-dimension (2D) Angle of Arrival (AOA) based precoding (beamforming) approach for multi-cell massive Multiple Input Multiple Output (m-MIMO) systems. The proposed approach enhances the performance of m-MIMO systems by overcoming pilot contamination that is caused by corrupted Channel State Information. It exploits the spatial properties of the propagated signals to distinguish between various users in the multi-cell m-MIMO system and selects the partial space for beamforming based on AOA information to suppress inter-cell and intra-cell interference. We mathematically analyze the impact of pilot contamination in the uplink training and determine the closed-form approximation of uplink Minimum Mean Square Error (MMSE) estimator. By using MMSE estimator as well as 2D AOA and channel estimation, a corresponding MMSE precoding which minimizes the effect of pilot contamination and improves the downlink achievable rate of the system is proposed. Furthermore, a comparison is made of the resolution between the various 2D AOA estimations for an m-MIMO and the optimum values closest bound for the proposed precoding was chosen to reduce pilot contamination. The performance of the proposed MMSE precoding based on 2D AOA estimation methods: 2D Unitary Estimation of Signal Parameters via Rotational Invariance Techniques (2D UESPRIT), 2D Fourier Domain Line Search MUSIC (2D FDLSM), and 2D Propagator Method (2D PM)) were evaluated and compared with pilot contaminated system and deterministic MMSE precoding in terms of their respective achievable rate. Simulation results reveal that the achievable sum rate gains of 2D UESPRIT, 2D FDLSM, and 2D PM based precoding which achieved 97.6%, 92.4%, 86.3% of the desired deterministic performance, respectively, compared to pilot contaminated system, which achieved only 90% even when in the best case of having only one contaminated user out of all users within the activated cell. Findings from our research indicate the feasibility of utilizing the AOA based beamforming to reduce pilot contamination effect that is inherent in a larger order m-MIMO system for the Fifth Generation (5G) wireless transmissions.

本文提出了一种新的预编码方法，该方法针对多小区大规模多输入多输出（m-MIMO）系统，利用了基于二维（2D）到达角（AOA）的预编码（波束成形）方法。所提出的方法通过克服由损坏的信道状态信息引起的导频污染来增强m-MIMO系统的性能。它利用传播信号的空间特性来区分多小区m-MIMO系统中的各个用户，并基于AOA信息选择用于波束成形的部分空间，以抑制小区间和小区间干扰。我们在数学上分析了上行链路训练中飞行员污染的影响，并确定了上行链路最小均方误差（MMSE）估计器的闭合形式近似值。通过使用MMSE估计器以及2D AOA和信道估计，提出了一种相应的MMSE预编码，该预编码将导频污染的影响降至最低，并提高了系统的下行链路可实现速率。此外，对用于m-MIMO的各种2D AOA估计之间的分辨率进行了比较，并为建议的预编码选择了最接近边界的最佳值以减少导频污染。提出的基于2D AOA估计方法的MMSE预编码的性能：通过旋转不变技术（2D UESPRIT），2D傅里叶域线搜索MUSIC（2D FDLSM）和2D传播器方法（2D PM）对信号参数进行2D单一估计。评估并与飞行员受污染的系统和确定性MMSE预编码进行了比较，它们分别是可达到的速率。仿真结果表明，与仅受污染的中试系统相比，基于2D UESPRIT，2D FDLSM和2D PM的预编码可实现的总比率增益分别达到了所需确定性的97.6％，92.4％，86.3％。即使在最佳情况下，在激活的单元格内所有用户中只有一个受污染的用户。我们研究的结果表明，利用基于AOA的波束成形来减少导频污染的可行性是第五代（5G）无线传输的较大阶m-MIMO系统固有的。即使在最好的情况下，即使在激活的小区内所有用户中只有一个受污染的用户，该指标也只能达到90％。我们的研究结果表明，利用基于AOA的波束成形来减少导频污染的可行性是第五代（5G）无线传输的较大阶m-MIMO系统固有的。即使在最好的情况下，即使在激活的小区内所有用户中只有一个受污染的用户，该指标也只能达到90％。我们的研究结果表明，利用基于AOA的波束成形来减少导频污染的可行性是第五代（5G）无线传输的较大阶m-MIMO系统固有的。