This paper considers a massive multiple-input multiple-output (MIMO) wireless uplink communication system over Rayleigh fading channels. In this system, two single-antenna nodes timely upload data to a base station (BS) with a large number of antennas on the same time-frequency resources. The small scale fading coefficients keep constant during two consecutive time slots, after which they change into new independent values. Assuming only the large-scale fading coefficients are available, we construct a maximum likelihood (ML) detector without the instantaneous small scale fading coefficients at the BS. The key idea is to utilize the absolutely additively uniquely decomposable constellation pair with a phase-shift keying (PSK) modulation scheme. Specifically, we use a rotated PSK constellation for one of the nodes to ensure the unique identification of transmitted symbols in a noise-free case. In a noisy case, we propose a power allocation framework for the proposed modulation scheme to improve the ML detector's average symbol error performance. Finally, simulations results verify the effectiveness of our proposed scheme.

本文考虑了瑞利衰落信道上的大规模多输入多输出（MIMO）无线上行链路通信系统。在该系统中，两个单天线节点及时将数据上载到具有相同时频资源的大量天线的基站（BS）。小规模衰落系数在两个连续的时隙中保持恒定，此后它们变为新的独立值。假设只有大范围的衰落系数可用，我们构造一个最大似然（ML）检测器，而无需在BS处获得瞬时的小范围的衰落系数。关键思想是利用具有相移键控（PSK）调制方案的绝对可加的，唯一可分解的星座对。特别，我们将旋转的PSK星座图用于其中一个节点，以确保在无噪声的情况下唯一识别传输符号。在嘈杂的情况下，我们为所提出的调制方案提出了一种功率分配框架，以改善ML检测器的平均符号错误性能。最后，仿真结果验证了我们提出的方案的有效性。