毫米波通信系统中的信道估计及峰均比压缩方法研究

发布时间：2018-03-27 07:24

本文选题：毫米波通信　切入点：大规模天线阵列　出处：《东南大学》2016年博士论文

【摘要】：随着第四代移动通信系统在全球范围内的普及,下一代移动通信系统的设计面临着频谱资源稀缺的现实问题。毫米波通信依靠丰富的频谱资源在下一代移动通信系统的相关研究中受到了广泛的关注。近年来,毫米波通信技术进入了一个快速发展的时期,针对一些技术性难题,研究人员提出了各种解决方案。在这一过程中,结合大规模天线阵列和波束成形的毫米波移动通信系统设计框架被不断地强化。在此基础上,本论文围绕毫米波信道估计和峰均比压缩方法进行了深入的研究。论文的主要贡献如下：针对大规模天线阵列与波束成形相结合的毫米波通信系统,本文提出了一种新的波束域毫米波信道估计方法。根据射线追踪(Ray-Tracing)模型,首先将毫米波信道矩阵估计问题转化为关于路径发射角、到达角及路径增益的参数估计问题。之后,利用波束域二维多重信号分类(MUSIC, MUltiple SIgnal Classification)空间谱和最小二乘法(Least-Squares)分别估计路径的方向角和增益系数,给出了毫米波参数信道估计算法。最后,对所提方案的可行性进行了论证,重点分析了波束成形对于波束域二维MUSIC空间谱模糊效应的影响,并给出了避免空间谱模糊效应且最大化可分辨路径方向角数量的波束成形设计充分条件。仿真结果表明,与目前基于波束训练及稀疏恢复的毫米波信道估计方法相比,所提出的信道估计方法具有更低的均方误差。为降低大规模天线阵列下通信接收机的硬件成本,本文还提出了一种新的基于低精度(1比特)模数转换器(ADC, Analog-to-Digital Converter)的发射端信道估计方法。量化精度的降低不仅能够抵消毫米波通信带宽剧增对于ADC功耗的影响,而且还有利于降低反馈开销,但是它同时给信道估计问题提出了巨大的挑战。针对这一问题,本文首先介绍了基于1比特扰动量化(DQ,Dithered Quantization)的信道估计方法,然后提出了基于1比特过水平采样(LTS, Level-Triggered Sampling)的信道估计方法。最后,重点分析了所提方法的渐进统计特性,证明了基于1比特LTS信道估计的一致性、渐进无偏性及渐进正态性。仿真结果表明,和基于1比特DQ的信道估计方法相比,基于1比特LTS的信道估计性能更接近于全精度采样情形下的理想结果。以上信道估计方法为获取毫米波信道状态信息(CSI, Channel State Information)提供了可行的备选方案。假设CSI已知,为提升发射端功率放大器(PA, Power Amplifier)的能量转换效率,本文提出了与波束成形相结合的峰均比(PAPR, Peak-to-Average Power Ratio)压缩优化方法。首先,在基于天线子阵列的波束成形方案的基础上,定义了发射端单个PA上的输入信号峰均比。然后分别建立了单天线峰值功率和预处理串扰约束的峰均比压缩优化问题,并推导了所得峰均比压缩方案的平均PAPR上界。最后,论文给出了求解以上峰均比压缩问题的优化算法。所提优化算法的核心步骤均涉及求解一类近端优化(Proximal Problem)子问题,论文通过验证Karush-Kuhn-Tucker条件,证明了所提算法求解该问题的最优性。功率削波影响下的仿真结果表明,和迫零预编码方案相比,本文提出的峰均比压缩方法能够获得更优的峰均比和误比特率性能。
[Abstract]:With the popularity of the fourth generation mobile communication system in the global scope, the design of the next generation mobile communication system is facing the reality of scarce spectrum resources of millimeter wave communication. Rely on the related research of rich spectrum resources in the next generation mobile communication system has received wide attention. In recent years, millimeter wave communication technology into a period of rapid development, aiming at some technical problems, researchers have proposed various solutions. In this process, combined with large-scale antenna array beamforming and millimeter wave communication system design framework was constantly reinforced. On this basis, the millimeter wave channel estimation and compression method of PAPR further research in this thesis. The main contributions of this thesis are as follows: according to the millimeter wave communication system with large array antenna beam forming combination, this paper proposes a new Millimeter wave beam domain channel estimation method. Based on ray tracing (Ray-Tracing) model, the millimeter wave channel matrix estimation problem is transformed into a path of emission angle, angle and parameters to the path gain estimation. Then, by using the beam domain two-dimensional multiple signal classification (MUSIC MUltiple, SIgnal Classification) and the least squares method (Least-Squares spatial spectrum) respectively to estimate path angle and the direction of the gain coefficient, given the parameters of millimeter wave channel estimation algorithm. Finally, the feasibility of the proposed scheme are discussed, analyzed the influence of spectrum beamforming for fuzzy effect beam domain two-dimensional MUSIC space, and are given to avoid the spatial spectrum and the effect of fuzzy maximum beam path resolution the number of direction angle forming design of sufficient conditions. The simulation results show that based on the current estimation of millimeter wave channel beam and sparse recovery training In comparison, the MSE of the proposed method has a lower channel. In order to reduce the communication receiver of large-scale antenna array under the hardware cost, this thesis proposes a new method based on low precision (1 bit analog-to-digital converter (ADC), Analog-to-Digital, Converter) of the transmitter channel estimation method not only reduce the quantization precision. To offset the effects of millimeter wave communication bandwidth increase for ADC power consumption, but also to reduce the feedback overhead, but it also challenges the channel estimation problem. To solve this problem, this paper introduces 1 bit quantization based on disturbance (DQ, Dithered Quantization) channel estimation methods, and then proposes 1 bit horizontal sampling based on (LTS, Level-Triggered Sampling) channel estimation method. Finally, focus on the analysis of the asymptotic statistical properties of the proposed method, proved that the 1 bit LTS channel based on Channel estimator is consistent, asymptotically unbiased and asymptotically normal. The simulation results show that the 1 bit DQ and based on the channel estimation method compared to 1 bit LTS channel estimation performance closer to the ideal results under the condition of full precision sampling. Based on above channel estimation method for millimeter wave channel state information (CSI Channel, State Information) provides a viable alternatives. Suppose CSI is known to enhance the transmitter power amplifier (PA, Power, Amplifier) of the energy conversion efficiency, this paper puts forward the combination and beamforming of PAPR (PAPR Peak-to-Average, Power Ratio) compression optimization method. Firstly, based on the antenna beamforming scheme the sub array based on the definition of the input signal emission peak end on a single PA than. And then set up the single antenna peak power and pretreatment with crosstalk constraint of PAPR reduction and optimization problem. Is the PAPR of the average upper bound of the PAPR compression scheme. Finally, the paper gives the PAPR optimization algorithm to solve the problem. The core compression step optimization algorithm involves solving a class of optimization of the proximal (Proximal Problem) sub problems, through the verification of Karush-Kuhn-Tucker conditions, proved the optimality of the proposed algorithm to solve this problem. Under the influence of wave power cutting simulation results show that compared with the zero forcing pre encoding scheme, the proposed PAPR compression method can obtain better PAPR and BER performance.

【学位授予单位】：东南大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TN928

【相似文献】