【作者】 董伟；

【导师】 李建东；

【作者基本信息】 西安电子科技大学 ， 通信与信息系统， 2008， 博士

【摘要】 随着实时多媒体通信、高速Internet接入等无线数据业务的发展,提高通信系统的速率和频谱利用率已成为急待解决的问题。在无线通信系统中,多输入多输出(Multi-input and Multi-output, MIMO)技术显著地提高信道容量而成为未来移动通信系统的热点研究技术之一。同时MIMO技术是一项复杂的技术,其研究内容包括了信道容量、空时编码、参数估计技术、检测技术等等。本文针对MIMO系统的参数估计技术和信号检测技术做了深入研究,主要的研究内容和创新性成果如下:1.研究了相关MIMO信道下的联合频偏和信道估计问题,由于MIMO信道存在不可避免的各种相关性,比如天线间的空域相关性和多径间的相关性,本文考虑信道是频率选择性的且具有空域相关性和多径相关性。根据贝叶斯原理提出了一种最大似然频偏估计器,推导了频偏估计的均值和方差,为了评估频偏估计器的性能,进一步推导了频偏估计的克拉美罗下界(Cramer-Rao Lower Bound, CRLB),证明了所提出的频偏估计器是渐近最优的。根据频偏估计值,推导了最小均方误差信道估计器,分析了频偏估计误差对信道估计器的性能影响。2.研究了MIMO系统中存在定时误差下的联合频偏和信道估计问题。首先推导了两个包含频偏和定时误差的等效系统模型,然后基于这两种系统模型提出了一种联合定时误差、频偏和信道的估计方法。该估计方法包括两步:第一步是根据第二个系统模型,推导了一种有效的最大似然频偏估计器;第二步是根据所获得的频偏估计值,利用第一个系统模型,将定时误差和信道估计表示成一个复合假设检验问题,然后再利用复合假设检验方法进行定时误差和信道估计。最后通过仿真验证了所提出的联合估计方法的有效性。3.研究了不同发送接收天线对之间存在不同频偏的MIMO系统的联合频偏和信道估计,给出了两种解决方法。一种是基于MUSIC和ML算法的估计方法,另一种是基于粒子群优化的估计方法。在第一种方法中,首先使用MUSIC方法估计出多个发射天线到某一接收天线的频偏子集,然后利用最大似然方法在这个有限子集中分离出不同天线对之间的频偏,最后在频率同步的基础上利用最大似然估计器对信道增益进行估计。在第二种方法中,首先根据粒子群优化理论估计出多个发射天线到某一接收天线的频偏,然后再利用最大似然估计器对信道增益进行估计。分析和仿真表明,方法一是次优的,而方法二是渐近最优的。4.研究了垂直分层空时系统(Vertical Bell-labs Layered Space-time, V-BLAST)中信号检测问题,将离散粒子群优化(Discrete Particle Swarm Optimization, DPSO)算法应用到垂直分层空时系统的检测中,提出了一种离散粒子群检测算法。针对DPSO检测算法有可能出现早熟现象,进一步提出了一种混合离散粒子群(Hybrid Discrete Particle Swarm Optimization, HDPSO)检测算法。HDPSO检测算法对DPSO检测算法的进化方程进行了重新设计,在搜索中以一定变异概率对选中的粒子进行变异,进一步改善了DPSO检测算法的性能。分析和仿真结果表明,所提出的算法与最优检测算法相比有更低的计算复杂度,与次优检测算法相比具有更好的误码率性能,为寻求新的V-BLAST系统检测算法提供了思路。5.为了进一步改善混合离散粒子群检测算法的性能,在HDPSO检测算法的基础上,结合并行干扰抵消(parallel interference cancellation, PIC)算法的快速局部收敛的优点,设计了两种垂直分层空时系统检测方法。方法1是使用HDPSO作为PIC的初始阶段,给后面干扰抵消阶段的PIC提供一个好的初始解,从而改善PIC算法的性能;方法2是将PIC嵌入到HDPSO的每一代中,选用一部分或全部粒子的位置矢量通过PIC进行局部同步寻优更新,来进一步改善种群的适应度值。将HDPSO和PIC的有机结合可以加快HDPSO检测算法的收敛速度,并且改善它的检测性能。更多还原

【Abstract】 Multi-input and multi-output (MIMO) technology is an important means to improve the performance of high-speed wireless communications, such as multi-media communications and wireless internet access. It is well known that over multipath fading channels, a MIMO system has much higher spectral efficiency than a conventional single-input and single-output (SISO) system. MIMO technology offers a variety of research fields, including channel capacity, space-time coding, parameter estimation, signal detection, and so on. This dissertation deals with the problem of parameter estimation and signal detection, its main contributions are as follows.1. The problem of joint frequency offset and channel estimation is studied for correlated MIMO channels. The channels are assumed to be frequency-selective and block fading with both spatial and multi-path correlations. A maximum-likelihood (ML) frequency offset (FO) estimator is proposed by using the Bayesian approach. The mean and variance of the FO estimation are derived. To evaluate the performance of the FO estimator, its Cramer-Rao low bound (CRLB) is also developed. It is shown that the proposed FO estimator is asymptotically optimal. Based on the FO estimate, we derive the linear minimum mean square error (LMMSE) channel estimator and analytically investigate the effect of frequency offset estimation error on the mean square error (MSE) performance of the channel estimator.2. The problem of joint frequency offset and channel estimation is studied for MIMO systems in the presence of timing error. Two equivalent signal models with frequency offset and timing error are given, and then a joint estimation method is derived. The proposed estimation method consists of two steps. Firstly, a ML frequency offset estimator is proposed based on the second signal model. Secondly, based on the FO estimate, we formulate the timing error and channel estimation as a problem of composite hypothesis testing according to the first signal model, and then solve the problem by the composite hypothesis testing approaches. Simulation results are performed to show the effectiveness of the proposed method.3. The problem of frequency offsets and channel estimation is studied for a MIMO system in flat-fading channels, where the frequency offsets are possibly different for each transmit antenna is considered. Two estimation methods are given. The first one is based on the multiple signal classification (MUSIC) and the ML algorithms. This estimation method has three steps. A subset of frequency offsets is first estimated with the MUSIC algorithm. Then all frequency offsets in the subset are identified with the ML algorithm. Finally channel gains are estimated with the ML estimator. The second one is based the particle swarm optimization theory. This estimation method has two steps. Frequency offsets are first estimated by the particle swarm optimization theory. Then channel gains are estimated by the ML estimator. Theoretical analyses and simulation results show that the first estimation method is suboptimal and that the second one is asymptotically optimal.4. The problem of signal detection is studied for vertical Bell-labs layered space-time (V-BLAST) systems. Firstly, a discrete particle swarm optimization (DPSO) detection algorithm is proposed by applying discrete particle swarm optimization to the signal detection of the V-BLAST system. Secondly, aiming at solving the premature convergence problem in DPSO detection algorithm, another detection algorithm, named by hybrid discrete particle swarm optimization (HDPSO) detection algorithm, is proposed. The HDPSO detection algorithm can be obtained by redesigning the evolution equation of the DPSO detection algorithm and introducing the mutation operator of the genetic algorithm, which improves the performance of the DPSO detection algorithm. Analyses and simulation results show that the proposed detection algorithms have lower computational complexity than the optimal detection algorithm and better detection performance than the suboptimal detection algorithms, to find a new method to solve the detection problem in V-BLAST systems.5. In order to further improve the performance of the HDPSO detection algorithm, two detection approaches that employ the HDPSO and parallel interference cancellation (PIC) algorithms in V-BLAST system are proposed. One of approaches is that the HDPSO algorithm is used as the first stage of the PIC to provide a good initial point for successive stages of the PIC, the other is that PIC is embedded into the HDPSO algorithm to further improve the fitness of the population at each generation. Such a hybridization of the HDPSO with the PIC can speed up its convergence. In addition, a better initial data estimate supplied by the HDPSO algorithm improves the performance of the PIC, and the embedded PIC improves the performance of the HDPSO.更多还原