【作者】 周志刚；

【导师】 程时昕；

【作者基本信息】 东南大学 ， 通信与信息系统， 2005， 博士

【摘要】 未来宽带无线通信系统面临着多径衰落信道影响和系统带宽效率等诸多问题的挑战。正交频分复用(Orthogonal Frequency Division Multiplexing-OFDM)技术通过将频率选择性多径衰落信道在频域内转换为平坦信道,可有效地对抗多径衰落的影响。而多天线(Multiple Input Multiple Output-MIMO)技术能够在空间形成独立的并行子信道同时传输多路数据流,有效地提高了系统的传输速率,在不增加系统带宽和改变系统功率要求的情况下增加频谱效率。因此,结合两者优势的MIMO-OFDM技术具备频谱利用率高、带宽可扩展性强等一系列特点,是下一代移动通信(Beyond 3rd Generation-B3G)系统有竞争力的候选方案。B3G系统高数据传输速率、高传输质量和移动性的要求对MIMO-OFDM系统的传输和检测技术提出了非常严格的要求,针对上述问题,本论文重点研究了MIMO-OFDM系统的导频序列设计、信道估计和信号检测等技术。论文针对MIMO-OFDM系统中的导频序列设计和信道估计问题,分析了多径信道中MIMO-OFDM系统传输的基带数学模型,推导了接收机的极大似然(ML)和最小二乘(LS)信道估计方法及其均方误差性能,并据此推导了MIMO-OFDM系统最优频域导频序列设计准则,推广传统两天线正交导频设计方法到了任意天线时的设计,并提出了一种导频信号峰均比为1,能达到正交导频时信道估计MSE下界的频域导频序列设计方法。并利用仿真评估分析提出的方法和传统设计方法在不同序列长度和天线数目情况下的性能差异。MIMO-OFDM系统中接收机检测算法的复杂度和性能优劣对系统实施和整体性能有很大影响,论文在分析传统检测算法优缺点的基础上,利用汉明子空间方法,提出了一种减少复杂度的软输出极大似然检测算法。MIMO-OFDM系统的接收检测可在各子载波上单独进行,最优序列(MLSE)和符号(MAP)检测在多天线和高阶调制情况下往往复杂度很高,而迫零(ZF)最小均方误差(MMSE)等传统线性检测算法复杂度低但性能相对较差,应用受到局限。本论文利用传统软输出的线性检测器生成信息比特的初始对数似然值,在此初始值形成的汉明子空间中搜索从而获得可靠度更高的比特对数似然软输出。由于生成的汉明子空间相对全空间大为减小且具有很高的可靠性,提出的算法可有效地减少了传统软输出的极大似然序列搜索算法的复杂度并能逼近极大似然性能。从最优接收理论而言,MIMO-OFDM系统中基带接收机的多模块联合极大似然接收可以取得最佳性能,但其复杂度使得在实际系统中往往难于实现。次最优的迭代检测译码算法在性能和复杂度上可以取得折衷,并有效地逼近系统容量性能。论文分析MIMO-OFDM系统中传统迭代检测译码算法优缺点,推导了基于MMSE的迭代检测译码算法,并在第四章中的减少复杂度的极大似然算法基础上改进,使其可以在子空间进行软输入软输出(SISO)搜索检测,并根据检测器需要进行比特硬判的特点改进译码器的软输出以及成员译码器的输入,从而提出了一种低复杂度的迭代检测译码算法。此算法可以有效地避免迭代检测译码过程中的多次求逆问题,复杂度分析和仿真评估也验证其优异的性能。移动通信系统的闭环传输可以获得更好的误码性能和更高的系统吞吐量,在分析了MIMO-OFDM系统慢变衰落信道环境下闭环传输技术的基础上,论文考察功率分配、预编码以及信道估计和反馈误差对闭环传输的影响,并提出了一种优化的发送预处理方案。首先在系统理想信道估计、理想信道反馈、发送功率均匀分配的情况下,针对ML、ZF、MMSE等接收机推导了最优发送预处理方案。然而实际系统中非理想信道估计和反馈信道的非理想会降低系统容量,针对这种情况,考虑系统信道估计和反馈信道的非理想,从MIMO-OFDM系统容量最大的角度推导最优发送预处理,并利用了内点算法优化系统的发送功率分配。推导的闭环发送预处理包括了预编码和功率分配,并考虑了信道估计和反馈信道误差方差的影响。提出的预处理方案可有效地降低由于非理想信道估计和反馈造成的容量平台效应,并提高系统的误比特率性能。在低信噪比和信道估计方差较小时,提出的优化功率分配的预编码方案可以获得高于理想信道估计情况下传统开环发送系统容量的性能更多还原

【Abstract】 The dispersive of wireless channel and the system bandwidth efficiency are the big challenges the future broadband wireless communication facing. Orthogonal frequency division multiplexing (OFDM) can effectively combat the multipath fading channel and convert the frequency selective channel into flat channel in frequency domain. Multiple input multiple output (MIMO) transmit multiple data streams in parrallel subchannels in spatial, thus greatly increasing the system throughput without additional bandwidth and power requirements. Combining the merits of two technologies, the bandwidth efficiency and expandability leads MIMO-OFDM to be a potential candidate for beyond 3rd generation (B3G) systems. The requirements for B3G systems such as high data rate, high quality and mobility make the transmission and receive techniques more strict. To solve the problems mentioned above, in this thesis, we focus on the design of the pilot sequences, channel estimation and signal detection techniques for MIMO-OFDM system.In order to design the pilot sequence and channel estimation algorithm, the baseband mathmatics model of MIMO-OFDM transmission in multipath fading channel is analyzed, and derived the channel estimation method and mean square error (MSE) performance based on the maximum likelihood (ML) and least square (LS). Based on these results, we derived the optimum criteria of frequency domain pilot sequences in MIMO-OFDM systems, and extend the traditional design method of orthogonal pilot sequences in two transmit antennas scenario to arbitrary transmit antennas scenario. We have also proposed a design method of frequency domain orthogonal pilot sequences which can achieve the lower bound of MSE and keep the PAPR of pilot signal to one. Finally their performance under different length of pilot sequence and number of transmit antennas are compared with traditional schemes and analyzed.The implementation and performance aspects of MIMO-OFDM system are usually depend on the complexity and performance of the receiver. We analyzed the merits and dismerits of traditional detector, and proposed a complexity reduced soft output maximum likelihood detector based on the hamming space theory. In MIMO-OFDM system, the detection can be performed on the each subcarrier, the optimum receive algorithm, maximum likelihood sequence estimation (MLSE) and maximum a posteriori (MAP) detector, are usually have much high complexity in multiple antennas and high order modulation scenario. However, the low complexity zero forcing (ZF) and minimum mean square error (MMSE) algorithm have less satisfied performance and limit its applications. In the proposed algorithm, we utilize the conventional linear detector at the initial stage to generate the log likelihood ratio of each code bits, then a hamming subspace can be spanned via the hard decision of these initial soft values. Themaximum likelihood search performed in this reduced space to generate the improved soft output of detector. For that this much smaller hamming space comared with the whole solution space has high initial reliability, the proposed algorithm can greatly reduce the complexity of maximum likelihood search and has the performance close to that of soft output MLSE.Base on the optimum receiver theory, the joint maximum likelihood receive of the whole receiver including detect, deinterleaving, decoding, etc has the best performance, while the high complexity made it can not be implemented in hardware currently and applied to the real systems. The iterative detect and decoding algorithm is a suboptimum joint receiver, can achieve the good trade off between the performance and complexity, and closing the Shannon limit performace. We analyzed the merits and dismerits of traditional iterative receiver, derived the iterative detect and decoing algorithm based on MMSE criterion, and modified the proposed complexity reduced maximum likelihood detect algorithm in 4th chapter, made it to be a low complexity soft input and soft output detect algorithm. Besides, we also modified the traditional soft input and soft output decoder’s input and output on the requirements of the detector. All this constitutes a novel low complexity iterative detect and decoding algorithm. This algorithm can avoid the multiple inversion operation in the traditional iterative algorithm based on MMSE and ZF. The analysis and computer simulation also demonstrated the advantages of proposed algorithm.The close loop transmission of wireless communication usually achieves good BER performance and much higher system throughput. Based on the analysis of close loop transmission under slow varing fading channel, we investigate the effects of the power allocation, precoding, channel estimation error and channel feedback error on the system, and proposed optimized transmit preprocessing scheme for MIMO-OFDM system. Firstly, under the assumption of perfect channel estimation, perfect feedback and equal power allocation between antennas and subcarriers, we derived optimal transmit preprocessing scheme for receiver based on ML, ZF, MMSE algorithm. However, the channel estimation and feedback can not be perfect in real world, we consided this condition and derived the optimal transmit preprocessing to maximum the system capacity, and optimize the power allocation between the antennas and subcarriers according to the inner point algorithm. The proposed preprocessing take into account a variance from the channel estimation and feedback, include precoding and power allocation jointly. The proposed scheme can effectively reduce the capacity floor made from the imperfect channel estimation and feedback, and improve the BER performance. When the variance is relative small and in low SNR region, the proposed precoding with optimized power allocation can achieve higher capacity performance compared with traditional open loop transmission under perfect channel estimation and feedback.更多还原