近几年来，随着移动互联网技术的普及以及智能无线终端的迅猛发展，人们对无线通信数据访问需求有了爆炸性的增长，这将促使蜂窝通信技术从当前的第四代移动通信系统(4G)快速演进到未来的第五代移动通信系统(5G)。目前针对5G乃至5G之后系统的技术方案论证已经在工业界和学术界全面铺开。其中，作为5G系统主要的备选技术，大规模多天线技术已经吸引了越来越多的关注与研究。大规模多天线技术旨在通过上百个(乃至更多的)基站天线单元来大幅度提升移动通信系统的吞吐量。为了充分地探索大规模多天线系统的潜力，解决实际应用中面临的挑战，仍然存在大量的问题值得研究。本文主要研究了大规模多天线系统的传输方案设计与系统性能分析问题。论文的主要贡献包括以下几个方面的工作： 1.本文提出了一种基于数据辅助的信道估计与信号检测方法来提升大规模多天线系统信道估计的精度。根据基站之间协作信息的不同，所提出的数据辅助的方法又细分成三种方案：(1)基站间无用户信息共享；(2)基站间共享干扰用户的导频信息；(3)基站间共享干扰用户的导频信息和已检测到的数据。在第一种方案中，目标基站将其服务用户已检测出的数据符号作为已知信息，与常规导频一起构成等效的长导频序列用来估计信道。在第二种方案中，邻近小区干扰用户的导频序列通过回程链路共享给目标基站用来估计干扰用户的信道，消除邻近小区的强干扰。在第三种方案中，邻近小区已检测出的强干扰用户的数据符号也通过回程链路共享给目标基站，目标基站通过数据辅助的信道估计方案可以获得对干扰用户更准确的信道估计，从而更彻底地消除来自邻近小区的干扰。为了对所提出的方案进行系统级性能分析，本文还提出了一种统一的架构用以分析任意确定小区形状的网络中传输信干噪比的分布。本文推导出了信干噪比分布的闭式表达式，并证明了所提方案可以显著地改善大规模多天线系统的性能。 2.本文提出了一种基于扩展导频的传输方案来提升大规模多天线系统中边缘用户的上行传输性能。根据用户到基站之间距离的大小，每个基站将其服务的用户分成边缘用户和中心用户，中心用户使用常规导频，边缘用户使用扩展了的长导频序列。目标基站先通过数据辅助的方式获得对小区中心用户精确的信道估计，再将中心用户的信号从总的接收信号中消除。通过基站之间导频信息的共享，目标基站联合估计出小区边缘用户与邻近干扰用户的信道。估计出的干扰用户信道可以在数据检测阶段抑制来自相邻小区的强干扰，从而极大地改善小区边缘用户的传输信于噪比。对于系统级性能分析，在考虑每个小区内用户位置随机分布的情况下，本文还推导出了常规方案与扩展导频方案的渐进信干噪比表达式，并进一步得到了信干噪比累计分布函数的闭式表达式。仿真结果证明所得闭式表达式与实际性能是一致的，所提出的方案可以大幅度地提高边缘用户的传输吞吐量。 3.本文利用数据辅助的信道估计方法来改善大规模多天线系统在时间相关信道中的上行传输性能。本文提出的方案把在上一帧中已经检测出的上行数据符号当作等效的导频信号，将其与信道时间相关系数一起用来抑制在当前帧的信道估计阶段来自相邻小区的干扰。进一步地，本文还推导出了在此方案下传输信干噪比的渐近线表达式，并验证了它的有效性。理论推导和仿真结果都证明了当信道时间相关系数值较大时，此方案可以在不改变传输帧结构的情况下，大幅改善大规模多天线系统中用户的传输信干噪比。 4.对于大规模多天线系统在时间相关信道中的下行传输，本文还提出了利用下行用户调度来改善系统性能。每个基站将其服务用户分成高移动性用户和低移动性用户。高移动性用户在数据辅助的信道估计以后立即进行下行传输，而系统中的各个基站协调地分配对低移动性用户的下行传输时延。此方案可以有效地抑制相关干扰导致的导频污染现象，每个小区中低移动性用户和高移动性用户的下行传输信干比都得到了明显的提升。本文还利用高斯近似的方法推导了下行传输信干噪比累计分布函数的闭式表达式，分析了传输信干噪比与多普勒频移的关系。 5.本文以有效吞吐量为性能指标，提出了一种统一的分析架构用于评估大规模多天线系统中软频分复用方案、软导频复用方案，以及频率与导频联合复用方案的性能。有效吞吐量在考虑了导频开销以及数据传输中断的情况下，计算单位时间内基站能够正确对目标用户传输的数据量。不同的方案之间存在着干扰抑制和传输效率之间的折中。考虑干扰功率的随机性，目标用户的信干噪比是一个随机变量。本文推导出了三种传输方案的渐进信干噪比的统一表达式，并在此基础上利用高斯近似进一步推导出了有效吞吐量的闭式表达式。闭式表达式与数值仿真结果是相吻合的，它可以直接用于计算各种方案对应的有效吞吐量占优的条件。 关键词：大规模多天线；方案设计；信道估计；性能分析 论文类型：应用基础

 In recent years, with the rapid development of mobile Internet and smart phones, the demand for wireless data traffic has grown explosively, which is promoting the evolution of cellular communication technology from the current fourth-generation mobile communication system (4G) to the future fifth-generation mobile communication system (5G). The technical solutions for 5G and beyond 5G have been widely investigated in both industry and academia. Massive MIMO, which is considered as a promising candidate of 5G, has gained significant attentions. Massive MIMO network can greatly increase the system throughput by developing hundreds of antennas (even more) at base stations. To fully explore the potential of massive MIMO network and solve its practical challenges, there are still a number of issues worth studying. Focusing on the transmission scheme design and system level performance analysis of massive MIMO network, the main contributions of this dissertation are summarized as follows. 1.A data-assisted uplink transmission framework is proposed for massive MIMO networks to relieve the pilot contamination. Specifically, three scenarios of inter-base-station connection are considered: (1) no pilot or detected data knowledge sharing; (2) pilot knowledge sharing via backhaul; (3) pilot knowledge and detected data sharing. In the first scenario, it is proposed to use the detected uplink symbols for the channel estimation refinement. In the second one, pilots of interfering users in close proximity are notified to the service base station for better interference mitigation. In the third one, pilots and the first few uplink data blocks of the closest interfering users are notified to the service base station, such that the interference from those interfering users is cancelled essentially. In order to obtain insights on system-level performance, a unified framework is established to analyze the signal-to-interference-and-noise ratio of the proposed schemes with arbitrary and fixed cell shapes and random user distribution. The closed-form expressions of the asymptotic performance bounds are derived, which demonstrate how much the aforementioned scenarios could help the uplink performance. 2.A novel pilot extension scheme is proposed for massive MIMO uplink transmission, and its performance is evaluated analytically. Specifically, the users of each sector are divided into two groups, namely near users and far users, according to their distances to the service base stations. It is proposed to schedule one more pilot sequence to each far user in the uplink frame, and share the pilot knowledge of strong interfering users via backhauls. Then the interference from the latter can be mitigated significantly. We also develop a general analysis framework for both conventional and proposed schemes, where the bounds on the cumulative distribution function of the uplink signal-to-interference ratio are derived instead of complex system-level simulations. The analytical bounds are tight and the proposed scheme could effectively improve the uplink signal-to-interference ratio for far users. 3.A data-assisted channel estimation scheme to exploit the temporal channel correlation in massive MIMO uplink transmission is proposed. Specifically, the previous decoded uplink frame is used to suppress the inter-cell interference in the channel estimation of the current frame. Based on the proposed scheme, we derive the asymptotic signal-to-interference expression, and the insights on how the network and channel parameters affect the system performance are obtained. It is shown numerically and analytically that the data-assisted scheme can reduce channel estimation errors and improve the performance of uplink massive MIMO systems significantly with the appropriate channel temporal correlation. 4.A novel downlink scheduling strategy for time-division duplexing massive MIMO networks is proposed to relieve the pilot contamination. Specifically, after data-assisted channel estimation, users of slow channel variation and simultaneous uplink pilot transmission are scheduled in different downlink subframes for adjacent cells, and hence the pilot contamination is signifi-cantly suppressed at the price of outdated channel state information. We investigate this trade-off by deriving the distribution of downlink signal-to-interference ratio in terms of Doppler effect, which can be approximated tightly and simply by Gaussian distribution. 5.A unified framework is proposed to analyze the downlink performance of massive MIMO networks with soft pilot reuse, soft frequency reuse, and joint frequency and pilot reuse, respectively. Different schemes have different trade-offs between interference mitigation and data transmission opportunities. Considering randomness of interference power, we derive the asymptotic closed-form expression of the average goodput, measuring the average number of bits successfully delivered to the users, for all reuse schemes as the performance comparison metric. KEY WORDS: Massive MIMO; Scheme Design; Channel Estimation; Performance Analysis TYPE OF DISSERTATION : Application Fundamental