【作者】 张雷；

【导师】 李少谦；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2009， 博士

【摘要】 在无线通信中，多天线传输(又名多入多出，Multi-Input Multi-Output，MIMO)技术能有效提升系统容量和改善误码率性能，已日益受到关注。实际系统的下行链路往往具有发射天线数较多、接收天线数较少(甚至为一)的非对称特点。若假设发射机完全未知信道信息，则传统的空间复用和空时编码等技术不能充分利用空间资源以进一步提升通信质量；反之，若假设发射机能获取理想信道信息，则可采用超强的预编码或波束成形等预处理技术以实现完全的分集和阵列增益，但这不符合无线通信的实际环境。因此，假设发射机能获取部分信道信息是合理且有意义的。另一方面，实际系统的整体等效信道在上下行链路并不满足互易性，故发射机获取信道信息的一种较为可行的方式是：接收机先估计信道信息，再将其反馈给发射机；同时，由于反馈链路的带宽受限和数字化实现需要，一般用若干比特映射而成的码本对信道信息量化之后再进行反馈。相比理想情形，此时预编码或波束成形的性能会有所损失，而且某些设计思路也有所不同。本文主要针对多天线系统中基于有限反馈的波束成形展开研究。对平坦衰落信道下的单用户系统，已有研究大多包含块衰落信道、准确信道估计或无反馈延迟等理想假设，第二章构建了关于“单用户时变多入单出(Multiple-Input Single-Output，MISO)信道中非理想反馈下的波束成形”的理论分析和系统设计框架。在建立非理想反馈系统模型的基础上，分析了方形或矩形QAM(Quadrature Amplitude Modulation)星座图的未编码误码率和遍历容量，并分别给出了误码率准确值和容量紧致上下界的闭合表达式。进一步推导了具有简洁表达式的误码率下界和容量更松下界，在给定归一化反馈比特数条件下，根据这两个下界得到了相同的最优帧长。对频率选择性信道，正交频分复用(Orthogonal Frequency Division Multiplexing，OFDM)技术将宽带信道转化为多路具有不同响应的窄带信道。若在每个子载波(子信道)独立反馈信道信息，则反馈量将与子载波数成正比，开销甚大。第三章研究了单用户有限反馈MISO-OFDM系统中最优波束成形向量的选取方法。在子载波分簇条件下，分别直接基于最小化QAM星座图的未编码误码率和最大化容量提出了簇内的最优波束成形向量选取准则，这两个准则分别在误码率和容量两个评估指标上优于其他已有方案。进一步基于簇内不同子载波信道频率响应之间的相关性提出了相应的简化次优算法，以较小的性能损失极大地降低了计算复杂度。最后将已有的自适应量化思想与所提簇内的波束成形向量选取准则相结合，给出了对应一个OFDM符号的完整反馈方案，它可在不影响性能的前提下进一步减小反馈量。第四章分析了多用户MIMO广播系统在固定反馈比特数条件下采用迫零波束成形时的和容量，给出了接入用户数等于发射天线数时其下界的闭合表达式，通过仿真比较了接入用户数小于发射天线数时不同参数条件下的和容量。更多还原

【Abstract】 In wireless communications multiple antenna transmission (viz. Multi-Input Multi-Output, MIMO) has received increasing attention in recent years, thanks to its potential for high spectral efficiency and improved quality of service. A predominant characteristic in a practical downlink is unsymmetrical structure of many transmit antennas and a few (even single) receive antennas. If no channel information is known to the transmitter, conventional spatial multiplexing or space-time coding can not explore space resource fully to improve performance. In contrast, if perfect channel information is available at the transmitter, some super precoding or beamforming techniques can be utilized to realize full diversity and array gains; however, this assumption is probably a utopia for most wireless environments. Consequently, multiple-antenna communications with partial channel information promise to have great practical value, because they are capable of offering the "jack of both trades." On the other hand, the total equivalent channel in practice is not strictly uplink-downlink reciprocal. So a reasonable approach for the transmitter obtaining channel information is to feedback the estimated one at the receiver. Furthermore, due to the rate constraint of the feedback link and requirement of digital realization, the channel information should be quantized by using some bits before feedback. Naturally, compared to the ideal case, precoding or beamforming with limited feedback has some loss in the performance and some different designs. This thesis focuses on beamforming with limited feedback in multiple antenna wireless systems.Most of prior research on beamforming in fiat fading channels includes several ideal conditions, such as block-fading channels, perfect channel information at the receiver, or error-free and zero-delay feedback link. We first study single user Multiple-Input Single-Output (MISO) systems over more realistic time-varying channels with imperfect feedback including estimation errors and feedback delay. An analytical framework is developed for evaluating average bit error rate (BER) of square/rectangle Quadrature Amplitude Modulation (QAM) constellations and ergodic capacity. Not only the close-form expressions of both exact BER and tighter capacity bounds are presented, but also the frame length is optimized according to the derived simple looser lower bound of BER/capacity from the viewpoint of system design.We further investigate beamforming in frequency-selective fading channels. By using Orthogonal Frequency Division Multiplexing (OFDM), the original broadband channels can be converted into many narrow-band subchannels (subcarreirs) with different frequency responses. We adopt rather subcarrier-clustering feedback than subcarrier-by-subcarrier one to reduce the immense overhead. Within one cluster, two optimal beamforming vector selection criteria are presented based on directly minimizing the average BER and maximizing the ergodic capacity, respectively. A simple sub-optimal algorithm is then proposed by exploiting the correlation of channel frequency responses at different subcarriers to reduce the computational complexity. The proposed schemes outperform other competitors in terms of BER or capacity. Furthermore, by integrating them into the existing adaptive quantization, a full feedback scheme corresponding to one OFDM symbol is developed to reduce feedback cost further without performance loss.We next extend our study to the multiuser case. The sum throughput is analyzed for multiuser MIMO broadcast systems using zero-forcing beamforming with limited feedback, and the close-form expression of its lower bound is given. We finally conclude the thesis by discussing some promising research directions.更多还原