Intelligent and Efficient Finite Alphabet Signal Processing Technologies for Future Wireless Communications

迄今为止，已经设想非正交多址（NOMA）和大规模MIMO（MaMIMO）是未来无线通信的两种主要潜在技术，因为它们满足未来无线通信的三个主要要求，即，巨大的数据速率、海量连接和低延迟。目前NOMA的研究主要集中在高斯输入或有限字母表（FA）输入的信道容量。然而，用于FA信号的方法是具体数值的。因此，可以从所获得的解决方案中得出对最佳和星座与每个单独用户星座之间的关系的有限洞察。另一方面，目前对MaMIMO的研究表明，显著影响其性能的主要因素之一是导频污染，即使当BS天线的数量变为无穷大时，导频污染也不会消失。因此，提出了用于MaMIMO的非相干空时码设计。然而，目前所有的非相干大规模空时码（MASTC）设计都存在严重的错误平层现象，即使信噪比达到无穷大也不会消失。此外，被动窃听者也不难配备大规模阵列，从而潜在地和物理地恶化上行链路安全性。 * 为了解决所有上述重要问题，申请人的长期研究目标是开发先进的有限字母表信号处理技术（FAST），以从信号估计和检测理论的观点智能地和有效地管理具有多个基站的上行链路或下行链路的一般复杂无线网络拓扑的FA干扰。在接下来的五年内，申请人主要瞄准以下三个目标：（a）开发用于上行链路SISO和下行链路MISO系统的FA NOMA的新FAST，（B）开发用于FA上行链路MaMIMO系统的新FAST，以及（c）开发用于物理地保护FA NOMA-MaMIMO系统的新FAST。为了实现这三个目标，具体提出了以下四个技术途径，即：(i)FA信号的加性唯一分解理论用于设计四种最优和星座，（ii）FA MIMO信号的乘性唯一分解理论用于设计三种最优MASTC，（iii）FA上行链路MaMIMO系统的超矩形覆盖理论用于设计仅在接收机处已知大尺度信道时的最优MASTC，以及（iv）MIMO信道的块相等QRS分解理论，用于设计基于ML的块连续抵消检测的最优FA预编码器。** 因此，这项研究计划开发了全新的FAST，用于智能有效地管理多用户通信的FA干扰，从而实现NOMA-MaMIMO提供的所有主要优势。*****

To date, it has been envisioned that non-orthogonal multi-access (NOMA) and massive MIMO (MaMIMO) are two major potential technologies for future wireless communications, because they meet three major requirements for future wireless communications, i.e., huge data rates, massive connectivity and low latency. The current research on NOMA focuses mainly on the channel capacity for either Gaussian inputs or finite-alphabet (FA) inputs. However, the methods for the FA signals are specifically numerical. Hence, a limited insight into the relationship between the optimal sum constellation and each individual user constellation can be drawn from the obtained solutions. On the other hand, the current research on MaMIMO has indicated that one of major factors to significantly affect its performance is pilot contamination, never vanishing, even when the number of BS antennas becomes infinity. Hence, non-coherent space-time code designs were proposed for MaMIMO. However, all the current non-coherent massive space-time code (MASTC) designs suffer a severe error floor phenomenon, never disappearing, even when SNR goes to infinity. In addition, it is also not difficult for a passive eavesdropper be equipped with a massive array, thereby; potentially and physically deteriorate uplink security. ***To address all the aforementioned significant issues, the applicant's long-term research objective is to develop advanced finite alphabet signal-processing technologies (FAST) to intelligently and efficiently manage FA interference for a general complicated wireless network topology with the uplink or the downlink having multiple base stations from the standpoint of signal estimation and detection theory. Within the next five years, the applicant targets mainly on the following three objectives: (a) to develop new FAST for FA NOMA for uplink SISO and downlink MISO systems, (b) to develop new FAST for FA uplink MaMIMO systems, and (c) to develop new FAST for physically securing FA NOMA-MaMIMO systems. In order to achieve these three objectives, the following four technical approaches are specifically proposed, i.e., (i) additively unique decomposition theory of FA signals for the design of four kinds of optimal sum constellations, (ii) multiplicatively unique decomposition theory of FA MIMO signals for the design of three kinds of optimal MASTC, (iii) super-rectangular cover theory for FA uplink MaMIMO systems for the design of an optimal MASTC when the large-scale channel is known only at the receiver, and (iv) block-equal QRS decomposition theory of MIMO channels for the design of an optimal FA precoder with ML-based block successive cancellation detection. ******Therefore, this proposed research program develops completely novel FAST for intelligently and efficiently managing FA interference for multi-user communications so that all the major benefits offered by NOMA-MaMIMO can be realized. *****