毫米波大规模天线系统协同分区多址理论与关键技术

Due to small antenna size, wide available spectrum and the facilitation of dense network deployment, the efficient combination of millimeter wave communications with massive multi-input multi-output technology (Massive MIMO) has a great potential for future wireless communications. However, due to strong correlation among numerous antennas integrated closely in millimeter wave user terminals, high power consumption caused by millimeter wave RF chains, and imperfect characteristic of millimeter wave devices, the traditional spatial asymptotic orthogonality for massive MIMO is not valid due to the existence of inter-cell multi-user mixed interference, which leads to the performance degradation of space division multiple access and the loss of system throughout and energy efficiency. The profound theoretical breakthrough is required to solve those issues. Based on big data analysis, random matrix theory, optimization theory and compressive sensing theory, this project investigates the multi-cell coordinated space division multiple access theory and key technology under multi-dimensional constraints (e.g. mixed interference, limited resource and dynamic wireless environment) for millimeter-wave based massive MIMO wireless systems from three perspectives of fundamental theory, system architecture and practical implementation. Specifically, this project aims to solve three key scientific problems: 1) The fundamental theorem of spatial asymptotic orthogonality under the existence of inter-cell multi-user mixed inference; 2) Multi-cell coordinated digital/analog mixed space division multiple access methodology under the constraint of limited resource; 3) Robust and efficient signal space orthogonal separation under multiple constraints including imperfect characteristic of millimeter wave devices, dynamic wireless environment and high pilot overhead. The outputs from this project will improve the system throughout and energy efficiency significantly for millimeter-wave based massive MIMO wireless systems, and more importantly, build up the theoretical basis for seamless integration of millimeter wave communications with massive MIMO.

毫米波通信具有天线尺寸小、频谱资源丰富、利于小区密集组网等优点，其与大规模天线的结合有望广泛应用于未来无线通信场景。然而，由于毫米波用户终端存在密集天线强相关、射频链路高功耗、器件环境非理想等多重约束，造成多址小区间混合干扰，导致空间渐近正交性难以成立、空分多址性能恶化、系统可达数据率和能量效率大幅下降等问题，亟需深层次的理论突破来支撑。本项目围绕上述问题，以大数据、随机矩阵、最优化和压缩感知等理论为主要分析工具，从基础理论、系统架构和方法实现这三个层面出发，研究复杂干扰、资源受限和动态环境下毫米波大规模天线系统协同分区多址理论与关键技术。具体而言，通过研究1）多址小区混合干扰下的空间渐近正交性理论；2）资源约束下的多基站协同数模混合空分多址机制；3）多重受限条件下的鲁棒高效信号空间正交分割等关键科学问题，大幅提升系统可达数据率和能量效率，为毫米波通信和大规模天线的有机结合夯实理论基础。

毫米波通信具有天线尺寸小、频谱资源丰富、利于小区密集组网等优点，其与大规模天线结合有望广泛应用于未来无线通信场景。然而，由于毫米波用户终端存在密集天线强相关、射频链路高功耗、器件环境非理想等多重约束，造成多址小区间混合干扰，导致空间渐近正交性难以成立、空分多址性能恶化、系统可达数据率和能量效率大幅下降等问题，亟需深层次的理论突破来支撑。项目围绕上述问题，以大数据、随机矩阵、最优化和压缩感知等理论为主要分析工具，从基础理论、系统架构和方法实现这三个层面出发，研究复杂干扰、资源受限和动态环境下毫米波大规模天线系统协同分区多址理论与关键技术。具体而言，通过深入研究：1）多址小区混合干扰下的空间渐近正交性理论；2）资源约束下的多基站协同数模混合空分多址机制；3）多重受限条件下的鲁棒高效信号空间正交分割等关键科学问题，大幅提升系统可达数据率和能量效率，为毫米波通信和大规模天线的有机结合夯实理论基础。项目执行期间，提出①基于波束时分多址的多址小区混合干扰协同；②自适应子连接数模混合预编码；③基于可重构反射阵天线的数模混合预编码；④毫米波Full-Dimension MIMO全连接子阵列数模混合预编码；⑤基于软导频复用的边缘用户干扰抑制；⑥基于统计数据辅助的毫米波波束训练；⑦部分天线激活的波束赋形；⑧结构化非均匀平面阵列天线设计。在IEEE Transactions等期刊上发表SCI收录论文18篇，在ICC/Globecom等国际会议上发表EI收录论文6篇，申请国家发明专利5项，获2016年中国通信学会科学技术奖二等奖（自然科学类）、国际旗舰会议IEEE ICC2017最佳论文奖等荣誉，圆满完成了任务书规定的研究目标。