In this work, we study massive multiple-input multiple-output (MIMO) precoders optimizing power consumption while achieving the users' rate requirements. We first characterize analytically the solutions for narrowband and wideband systems minimizing the power amplifiers (PAs) consumption in low system load, where the per-antenna power constraints are not binding. After, we focus on the asymptotic wideband regime. The power consumed by the whole base station (BS) and the high-load scenario are then also investigated. We obtain simple solutions, and the optimal strategy in the asymptotic case reduces to finding the optimal number of active antennas, relying on known precoders among the active antennas. Numerical results show that large savings in power consumption are achievable in the narrowband system by employing antenna selection, while all antennas need to be activated in the wideband system when considering only the PAs consumption, and this implies lower savings. When considering the overall BS power consumption and a large number of subcarriers, we show that significant savings are achievable in the low-load regime by using a subset of the BS antennas. While optimization based on transmit power pushes to activate all antennas, optimization based on consumed power activates a number of antennas proportional to the load.

翻译：本文研究了大规模多输入多输出（MIMO）系统中在满足用户速率需求的同时优化功耗的预编码器设计。首先，我们解析地表征了在低系统负载下最小化功率放大器（PA）功耗的窄带与宽带系统解，此时每根天线的功率约束不具约束性。随后，我们聚焦于渐近宽带场景。进一步研究了整个基站（BS）的功耗以及高负载场景。我们获得了简单解，且在渐近情况下的最优策略简化为：基于已知的活跃天线之间的预编码器，寻找最优活跃天线数量。数值结果表明：在窄带系统中，通过天线选择可实现较大功耗节省；而在仅考虑PA功耗的宽带系统中，所有天线均需激活，因此节省幅度较小。当考虑整体基站功耗及大量子载波时，我们证明在低负载场景下通过使用部分基站天线可显著节省功耗。基于发射功率的优化倾向于激活所有天线，而基于总功耗的优化则激活与负载成正比的天线数量。