Full-duplex (FD) is a promising technology for increasing the spectral efficiency of next-generation wireless communication systems. A major technical challenge in enabling FD in a real network is to remove the self-interference (SI) caused by simultaneous transmission and reception at the transceiver, and the SI cancellation performance depends significantly on the estimation accuracy of the SI channel. In this study, we proposed a novel partial SI channel training method for minimizing the residual SI power for FD massive multiple-input multiple-output (MIMO) systems. Based on an SI channel training framework under a limited training overhead, using the proposed scheme, the BS estimates only a part of the SI channel vectors, while skipping the channel training for the other remaining SI channel vectors by using their last estimates. With this partial training framework, the proposed scheme finds the optimal partial SI channel training strategy for pilot allocation to minimize the expected residual SI power, considering the time-varying Rician fading channel model for the SI channel. Therefore, the proposed scheme can improve the sum-rate performance compared with other simple partial training schemes for FD massive MIMO systems under a limited training overhead. Numerical results confirm the effectiveness of the proposed scheme for FD massive MIMO systems compared with the full training scheme, as well as other partial training schemes.

中文翻译：

---

全双工大规模MIMO系统的小干扰信道训练

全双工（FD）是提高下一代无线通信系统频谱效率的有前途的技术。在实际网络中启用FD的主要技术挑战是消除由收发器同时发送和接收引起的自干扰（SI），SI消除性能在很大程度上取决于SI信道的估计精度。在这项研究中，我们提出了一种新颖的局部SI信道训练方法，用于最小化FD大规模多输入多输出（MIMO）系统的残留SI功率。基于在有限的训练开销下的SI信道训练框架，使用所提出的方案，BS仅估计一部分SI信道向量，而通过使用它们的最后估计来跳过其他剩余的SI信道向量的信道训练。在这种局部训练框架下，考虑到SI信道的时变Rician衰落信道模型，所提出的方案找到了用于导频分配的最佳局部SI信道训练策略，以最小化预期的残留SI功率。因此，与有限的训练开销下的FD大规模MIMO系统的其他简单的局部训练方案相比，所提出的方案可以提高求和率性能。数值结果证实了与大规模训练方案以及其他局部训练方案相比，该方法对于FD大规模MIMO系统的有效性。因此，与有限的训练开销下的FD大规模MIMO系统的其他简单的局部训练方案相比，所提出的方案可以提高求和率性能。数值结果证实了与大规模训练方案以及其他局部训练方案相比，该方法对于FD大规模MIMO系统的有效性。因此，与有限的训练开销下的FD大规模MIMO系统的其他简单的局部训练方案相比，所提出的方案可以提高求和率性能。数值结果证实了与大规模训练方案以及其他局部训练方案相比，该方法对于FD大规模MIMO系统的有效性。