随着移动互联网和物联网的快速发展，移动数据业务呈现爆炸式增长，海量终端连接到移动网络。相比于第四代(4G)无线通信系统，第五代(5G)无线通信系统的容量要有1000倍的提升。为此，提出了多种新型技术来提高5G无线通信系统的性能，毫米波、大规模天线、超密集网络是其中的三大关键技术。 毫米波因其巨大的通信带宽受到广泛的关注与研究，应用场景包括蜂窝网络、室内热点、无线回传与前传、设备到设备、车载网络等。同时，由于毫米波的频率很高，其信道传播具有如下特性：路径损耗大；穿透损耗大；易受氧气、雨水、植被衰减影响；高定向性；易受人体阻挡；多径稀疏性等。由于其波长短，天线的形状系数小，阵列天线的尺寸可以做到很小，毫米波可与大规模天线互补结合，充分发挥各自的优点，极大提升系统的性能。这些新的信道传播特性给毫米波的应用带来了极大的挑战。 信道测量与建模是评估无线通信系统性能的前提，5G毫米波通信系统的实现需要进行大量的毫米波信道测量，提出相应的毫米波信道模型，用于评估系统的性能。本文简要介绍了5G无线通信系统的发展，全面总结了毫米波的频谱分配、信道传播特性、信道测量方法、信道探测器、信道参数估计算法、信道建模方法与标准化信道模型，进行了大量的多频段多场景毫米波信道测量，建立了毫米波信道测量数据库，并提出了多种毫米波信道模型。完成的主要工作与创新性成果如下： ⑴60GHz室内信道测量与建模：基于矢量网络分析仪，搭建并扩展了毫米波信道测量系统，开展了 60GHz室内信道测量，证实了定向天线旋转与虚拟阵列扫描两种信道测量方法的一致性，揭示了不同天线配置对信道传播特性的影响；使用空间交替广义期望最大化(SAGE)数据后处理算法，提取信号的多径参数，建立了60GHz信道的射线追踪模型与SV模型。 (2)多频段毫米波大规模天线信道测量与建模：在室内环境下首次开展了11 GHz、16GHz、28GHz、38GHz4个频段的毫米波大规模天线信道测量，建立了毫米波信道测量数据库，使用SAGE算法提取子阵列的多径参数，研究了多径参数沿阵列轴的变化。比较了不同频段毫米波的信道传播特性，揭示了毫米波频段大规模天线配置下的近场球面波、簇的生灭与非平稳特性，进而提出了相应的传播图模型与几何随机信道模型。 (3)毫米波多天线三维宽带几何随机信道模型：提出了一种新颖的毫米波多天线三维宽带几何随机信道模型。通过三维空间中的齐次泊松点过程产生簇，通过分布在收发天线两端的双球产生与多径对应的散射体，建模多径在时延及角度域的成簇特性，同时考虑了大尺度路径损耗模型与直射径概率模型。通过与毫米波信道测量结果的对比，证实了模型的准确性。 (4)基于大数据与机器学习的无线信道模型：提出了一种新颖的基于大数据与机器学习的无线信道建模及统计特性预测方法。运用前向神经网络与径向基函数神经网络，预测了信道的接收功率、时延扩展、角度扩展等统计特性，同时使用信道测量数据与信道模型仿真数据进行训练与测试，证实了所提方法的性能。 综上所述，本文在毫米波信道探测器、信道测量、参数估计与信道建模等方面开展了相关的研究工作，进行了大量的多频段多场景毫米波信道测量，提出了多种毫米波信道模型，并通过信道测量予以验证，这些工作对于5G毫米波无线通信系统的实现具有重要的指导意义。 关键词：5G无线通信；毫米波信道；信道探测器；信道测量；信道建模

 With the rapid developments of mobile Internet and Intemet-of-things (IoT), mobile data services have grown explosively and enormous end users have connected to the mobile network. Compared with the fourth generation (4G) wireless communication systems, the capacity of the fifth generation (5G) wireless communication systems should have an increase of 1000 times. To realize it, many new technologies have been proposed to improve the performance of 5G wireless communication systems, among which millimeter wave (mmWave), massive multiple-input multiple-output (MMO), and ultra-dense network (UDN) have been the big three key technologies. Mm Wave has drawn many attentions and researches due to its huge communication bandwidths. Its application scenarios include cellular network, indoor hotspot, wireless backhaul and fronthaul, device-to-device (D2D), vehicular networks, etc. Meanwhile, mm Wave channel has some different propagation characteristics due to its high frequency, such as the high path loss, high penetration loss, effect of oxygen/water/vegetation attenuation, high directivity, human blockage, and multipath sparsity. Because its wavelength is short, the antenna form factor is small, the size of antenna array can be small and mm Wave can be complementary combined with massive MIMO to make full use of their advantages and largely improve the system performance. All these new channel propagation characteristics will bring great challenges to the applications of mmWave. Channel measurements and modeling are prerequisite for evaluation of wireless communication system performance. To realize and evaluate 5G mmWave communication systems, various mmWave channel measurements should be conducted and corresponding channel models should be proposed. This thesis gives a brief introduction of the developments of 5G wireless communication systems. Moreover, a comprehensive summary of mmWave frequency allocations, channel propagation characteristics, channel measurement methods, channel sounders, channel parameter estimation algorithms, channel modeling approaches, and standardized channel models is given. Extensive multi-frequency multi-scenario mmWave channel measurement campaigns are conducted and channel measurement data base is built. Various mmWave channel models are proposed. The main works and novelties are below: (1)60 GHz indoor channel measurements and modeling: A mm Wave channel measurement system is built and extended based on a vector network analyzer (VNA). The 60 GHz channel measurements are conducted in an indoor environment. The rotate directional antenna (RDA) and uniform virtual array (UVA) measurement methods are proved to be consistent. The effects of different antenna configurations on channel propagation characteristics are revealed. The space-alternating generalized expectation-maximization (SAGE) data post-processing algorithm is applied to extract multipath component (MPC) parameters. The ray tracing model and the SV model are developed. (2)Multi-frequency mm Wave massive MIMO channel measurements and modeling: MmWave massive MIMO channel measurements are conducted in an indoor environment at 11, 16, 28, and 38 GHz bands for the first time and channel measurement data base is built. The SAGE algorithm is used to extract MPC parameters of the sub-array measurement data and the variations of MPC parameters over the array are studied. Channel propagation characteristics of different mm Wave bands are compared. Massive MIMO properties such as near field spherical wavefront, cluster birth-death, and non-stationary properties at mm Wave bands are validated. The propagation graph model and geometry based stochastic model (GBSM) are proposed. (3)A mm Wave MIMO three dimensional (3D) wideband GBSM: A novel mm Wave MIMO 3D wideband GBSM is proposed. Clusters are generated by a homogeneous Poisson point process (PPP) in 3D space. Scatterers are distributed in the two spheres at transmitter (Tx) and receiver (Rx) sides and corresponded to rays. The two spheres are used to model the MPC clustering property in delay and angular domains. The large-scale path loss model and line-of-sight (LOS) probability model are both considered. The accuracy of the proposed model is verified by comparing with mmWave channel measurement results. (4)A big data and machine learning (ML) enabled wireless channel model: A novel big data and ML enabled wireless channel modeling and statistical properties prediction method is proposed. The feed-forward neural network (FNN) and radial basis function neural network (RBF-NN) are both applied to predict channel statistical properties such as received power, delay spread (DS), and angle spreads (ASs). The performance of the proposed method is validated by training and testing of datasets collected from both real channel measurements and GBSM simulations. In summary, this thesis has done many works on mmWave channel sounders, channel measurements, parameter estimations, and channel modeling. Extensive multi-frequency multi-scenario mmWave channel measurements have been conducted. Various mmWave channel models have been proposed and verified by channel measurements. These works can have an important guidance significance for the realization of 5G mmWave wireless communication systems. Keywords: 5G wireless communications; millimeter wave channel; channel sounder; channel measurements; channel modeling