Research On Key Technologies Of Multi-Antenna Multi-user Communication System

In the research on wireless communication systems, multiple antenna techniques have attracted extensive attention in recent years, due to their potential for high spectral efficiency and high flexibility. Although the application techniques for single user systems are well developed, it is still an open issue for exploiting the spatial multiplexing gain by using multiple antennas at the basestation (BS). In this dissertation, we investigate various realization strategies for multi-antenna multi-user systems, and analyze the uplink/downlink communication performance specificly. The main contributions of this thesis are as follows:Firstly, for the downlink of multiple antenna systems with the ability of multiuser spatial multiplexing, a novel precoding algorithm based on a series of linear constraints on jamming is proposed, and a proof on its equivalence to the linearly precoding schemes of block diagonalization (BD) and signal to leakage ratio (SLR) maximization is also given. Besides, closed-form expressions of the optimal precoding weights of the algorithm proposed are derived, therefore resulting in the remarkable reduction on the complexity of the existed schemes, by avoiding the operations of singular value decomposition (or generalized eigenvalue decomposition). Also, using the results above, we derive the exact probability density function (PDF) of the receiving SINR (SIR), and furthermore the analytical expressions of the performance evaluation metrics.Secondly, it is well known that the communication performance of multiple antenna systems highly depends on the amount of channel state information (CSI) known at the basestation. So, for the systems that the downlink CSI is obtained through feedback channel, we investage a precoding scheme based on the limited feedback quantization codebook. Here a minimum jamming precoding algorithm utilizing channel direction information only is derived, and then we present a limited feedback linear precoding strategy based on random vector quantization (RVQ) codebook. Compared with the conventional BD scheme, the algorithm proposed exhibits much higher performance. However, due to the effect of quantization noise mismatch, this improvement fluctuates with the increasing of the SNR, and eventually converges to the BD scheme. To eliminate the impact above, a robust precoding algorithm is also presented. Simulation results show that, this algorithm can ensure the advantage over the BD scheme, and at the same time, maintain the performance while the SNR increasing, therefore improve the robustness of the system.Thirdly, we study the multiuser selection strategy for the downlink linearly precoding systems. Here a scheme based on the generalized minimum jamming criterion is considered, and simulation results show that it can obtain better performance compard with the BD algorithm, and more importantly, it can surport much more users at the same time (frequency) slot. Based on the selection scheme above, its fast realization versions are also proposed. The performance of the fast selection algorithms is close to that of the exhaustive search, whereas the reduction in the system complexity is quite large.Lastly, we analyze the uplink performance of multiple antenna systems using receiver diversity techniques. Especially, we focus on the situation of the generalized distributed antenna system (GDAS), which can significantly reduce the impact of shadowing in the wireless fading channel. Because of the mathematical complexity of the composite fading in GDAS, here an approximation method is used to derive the PDF of the combining output SNR, and the high precise analytical expressions of outage probability and bit error ratio are also presented.