Spatial Characteristics And Capacity Evaluation Of Wideband MIMO Channels Based On Radio Propagation Measurements

Radio channel is the media of signal transmission in the wireless communication systems. The actual performance of the system highly depends on the channel characteristics. In order to optimize the system performance, the channel characteristics are required to be considered comprehensively when designing a wireless communication system. Channel measurement is the most straightforward way to obtain channel properties. A variety of channel behaviors can be observed and collected via channel measurement, and the key parameters characterizing the channel behaviors are able to be extracted, which further set up the foundation for the channel modeling and simulation. Based on extensive field measurements, the spatial characteristics and capacity evaluation of the wideband multiple-antenna channels are presented in the thesis. The work can be categorized as four parts:First, the improvement of space-alternating generalized expectation maximization (SAGE) algorithm, which is applied to channel parameter estimation, is proposed. On one hand, the initialization process is ameliorated. A concept of local area is defined as the range within several wavelengths. Considering the correlation between the adjacent snapshots, when extracting the channel parameters from the snapshots within a given local area, only the first snapshot should be initialized by the traditional successive interference cancellation (SIC) method, whereas for the other snapshots, the initial values of channel parameters are set as the corresponding results of the previous snapshot directly. On the other hand, the searching strategy is improved. Since the multiple-path components are commonly clustered and the paths belonging to a specific cluster share similar parameters, a concept of prior searching area is defined as the interregional within several searching steps of the extracted parameters. When updating a parameter, the algorithm searches the prior searching area firstly to find a value that makes the likelihood function monotonic non-decreasing and maximized. If the value exists, the parameter is updated to it, and if not, the algorithm searches elsewhere. The novel initialization process and searching strategy enhance the efficiency of the algorithm prominently.Second, the spatial channel parameters such as power angular spectrum (PAS), and circular angle spread (CAS) are proposed for suburban macro (SMa), urban macro (UMa), urban micro (UMi) and relay scenarios. In SMa scenario, the transmitter (Tx) and receiver (Rx) are always under line-of-sight (LoS) or obstructed line-of-sight (OLoS) condition. In this case, the transmitted power are centralized at the LoS direction, whereas the received power presents a visible two-path effect, in which one path comes from the LoS propagation or diffraction by rooftop, and the other is due to the one-bounce reflection. In UMa scenario, when Tx and Rx are under LoS condition, both transmitted and received power are concentrated at the LoS direction, but the local scattering effect are more conspicuous at Rx side. In NLoS condition, the radio waves mainly arrive at Rx via one-bounce diffraction of reflection. As for the UMi scenario and the relay station to mobile station (RS-MS) link in relay scenario, since the Tx antenna is lower than the buildings around, the radio waves can only propagate along the streets so that the streets presents a significant wave-guide effect, which leads to smaller CAS. In addition, for the base station to relay station (BS-RS) link in relay scenario, the major propagation mechanism in LoS condition is direct propagation, whereas is rooftop diffraction in NLoS condition. The angle spread of arrival is smaller than the reference value due to the higher antenna altitude.Third, the capacity gain and outage capacity corresponding to a variety of antenna configurations are evaluated in SMa, UMa and UMi scenarios. Totally 7 types of antenna configurations varying in element number, element spacing and polarization are considered. The numerical results indicate that for a given scenario, increasing the antenna number at only Tx side has little benefit on both capacity gain and 5% outage capacity. When the value of ASD is small, enlarging the antenna spacing improves the outage capacity to some extent, but hardly shows any influence on capacity gain. The dual polarized antennas perform better than the vertically polarized antennas. Furthermore, the performance of a certain antenna configuration in different scenarios reveals that the values of CAS have little impact on the capacity gain. However, the outage capacity grows with the increasing of CAS, especially when the signal to noise ratio (SNR) is low.Fourth, the ergodic capacity of amplify-and-forward (AF) relay channels with limited feedback is evaluated. A novel feedback scheme for channel eigenvalues is proposed, in which the minimum mean squared error (MMSE) condition is employed as the codebook design criterion and Lloyd algorithm is applied to the codebook generation. Then, for the 4×4×4 AF relay system, the values of ergodic capacity corresponding to various antenna configurations, precoding codebooks, feedback quantities and power allocations strategies are presented and compared. The evaluation results indicate that dual polarized antennas bring about higher capacity than vertically polarized antennas. When the feedback quantity is fixed, different precoding codebooks presents similar performance. For a given type of precoding codebook, more feedback bits lead to higher capacity, but a tradeoff is required between the quantization error and feedback overhead. Additionally, optimal power allocation significantly overperforms the uniform power allocation, thus the proposed feedback scheme for channel eigenvalues is indispensible.It can be concluded that the measurement data and analysis results presented by the thesis will be supportive in the network planning and technique evaluation of the future wireless communication systems.