Hybrid Precoding for Massive MIMO With Low Rank Channels: A Two-Stage User Scheduling Approach
Hybrid Precoding for Massive MIMO With Low Rank Channels: A Two-Stage User Scheduling Approach
复制标题
DOI：
10.1109/tcomm.2020.2991805
复制
发表时间：
2020-08-01
影响因子：
8.3
通讯作者：
Fusco, Vincent F.
中科院分区：
文献类型：
--
作者：
Almradi, Ahmed;Matthaiou, Michail;Fusco, Vincent F.
为了充分发挥大规模多输入多输出混合模拟和数字预编码在频分双工单小区系统中的优势，提出了一种基于信漏加噪声比度量的两级预编码器。该技术的主要思想是仅根据发射端长期的信道统计信息，即信道均值和重构降阶协方差统计量，联合设计模拟预编码器，而数字预编码器则是基于降维有效重构信道的瞬时信道状态信息来设计的。因此，我们可以显著地减少下行链路训练和上行链路反馈开销，类似于所得到的有效信道的等级。还研究了发射端全信道状态信息(CSIT)和统计CSIT的两个极端。全面CSIT和统计CSIT之间的绩效差距证实了拟议的两阶段CSIT方法的重要性。然后，这些预编码器被扩展到多小区系统。结果表明，数字基带预编码器设计问题归结为广义瑞利商问题，而模拟预编码器设计问题归结为商迹问题，也称为比率迹问题。这些降维问题通过广义特征值分解方法得到解决。最后，在只调度部分用户的多用户分集的情况下，为了显著减轻信道估计和反馈开销，提出了一种低复杂度的单级和两级CSIT联合用户调度和预编码器算法。
To fully reap the benefits of massive multiple-input multiple-output hybrid analog and digital precoding in frequency division duplexing single-cell systems, a two-stage precoder is developed utilizing the signal-to-leakage-plus-noise ratio metric. The main idea of this technique is to jointly design the analog precoder based only on the long-term channel statistics information at the transmitter, i.e., the channel mean and reconstructed reduced rank covariance statistics, while the digital precoder is designed based on the instantaneous channel state information of the reduced dimensionality effective reconstructed channel. Consequently, we can significantly reduce the downlink training and uplink feedback overhead analogously to the rank of the resultant effective channel. The two extremes of full channel state information at the transmitter (CSIT) and statistical CSIT are also investigated. The performance gap between the full and statistical CSIT corroborates the importance of the proposed two-stage CSIT approach. These precoders are then extended to multi-cell systems. It is shown that the digital baseband precoder design problem reduces to the generalized Rayleigh quotient problem, while the analog precoder design problem reduces to the quotient trace problem, also known as the ratio trace problem. These dimensionality reduction problems are solved via the generalized eigenvalue decomposition method. Finally, in the presence of multiuser diversity where only a subset of the users are scheduled, to considerably alleviate the channel estimation and feedback overhead burden, a low-complexity one-stage and two-stage CSIT joint user scheduler and precoder algorithms are developed.