【作者】 王志强；

【导师】 王军；

【作者基本信息】 电子科技大学 ， 通信与信息系统， 2022， 博士

【摘要】 在5G网络以及未来B5G/6G网络中,将采用毫米波频段提供的超大带宽满足其网络容量需求。然而,毫米波严重的路径损耗导致系统传输距离有限。利用大规模多入多出(Multiple Input Multiple Output,MIMO)天线进行波束成形提高增益,可以有效抵抗高路径损耗带来的影响。同时,结合中继协作传输技术可进一步增加通信传输距离。虽然毫米波大规模MIMO联合中继协作传输技术能够为未来无线通信需求提供有效支撑,但受限于当前器件工艺水平,毫米波大规模MIMO系统面临着高复杂度、功耗和成本等挑战,严重阻碍了系统的实用化进程。为应对上述挑战,模数混合阵列架构与低精度量化模数转换器(Analog-toDigital Converters,ADC)/数模转换器(Digital-to-Analog Converters,DAC)收发机等新型系统硬件架构应运而生。针对不同的应用场景,选择合适的硬件架构形式来降低系统复杂度是一种直接有效的解决方案。然而,采用低复杂度硬件架构以及大规模天线阵列的宽带非理想特性会导致系统性能下降。如何设计稳健的信号处理算法是一个值得研究的问题。此外,在中继协作系统中存在多个传输节点,使得系统优化设计更为复杂。基于此,本文从毫米波大规模MIMO系统低复杂度硬件架构形式出发,通过探索实际信道特性,设计高精度中继协作信道估计和稳健波束成形算法,以提升系统性能。本文的主要工作和贡献如下:(1)本文研究了模数混合阵列架构下的高效能重叠子阵连接结构。该结构是一种介于全连接结构和分子阵连接结构之间的可变子阵连接结构,能够在保证系统性能的同时降低移相器使用数量。基于重叠子阵连接结构,本文提出了一种低复杂度毫米波多用户下行链路模数混合波束成形算法,并详细分析了每个射频链路所连接的天线数目对系统性能的影响,结果表明,在稀疏毫米波信道下存在一个最佳的天线连接数目。该结果有助于指导实际模数混合阵列架构方案设计。(2)本文探索实际中继协作系统中信道的双时间尺度特性,设计了高精度的双时间尺度中继协作信道估计算法。利用实际系统中的中继-基站间上行信道变化比用户-中继间上行信道慢的物理特性,将信道估计问题建模成双时间尺度近似最大似然(Approximate Maximum Likelihood,AML)问题。本文基于块坐标下降方法(Block Coordinate Descent,BCD)提出了一种离线批处理(AML-Batch)算法求解上述优化问题。为克服AML-Batch算法在实时在线应用中面临的高复杂度以及高存储缺点,本文基于双阶段随机优化理论进一步提出了一种低复杂度在线学习(AML-Online)信道估计算法。仿真结果表明,本文所提出的双时间尺度中继信道估计算法通过利用多个时刻的导频数据,估计性能优于传统的单时间尺度信道估计算法。在此基础上,本文进一步研究了智能反射表面(Intelligent Reflecting Surface,IRS)辅助的毫米波协作通信系统级联信道估计问题,通过探索级联信道在时间尺度上的结构稀疏特性,本文提出了一种基于正交匹配追踪的递归在线(Recursive Structure Orthogonal Matching Pursuit,RS-OMP)双时间尺度稀疏级联信道估计算法,降低了导频开销并提升了估计精度。(3)针对低精度量化大规模MIMO双向中继系统稳健中继波束成形问题,本文基于结构化设计的分离最小均方误差(Minimum Mean Square Error,MMSE)方案,提出了一种稳健的MMSE(Robust MMSE,R-MMSE)中继波束成形算法。所提算法考虑了低精度量化ADCs/DACs收发机产生的量化噪声,从而能够在一定程度上减轻低精度量化ADCs/DACs收发机带来的性能损失,在大规模用户接入场景下性能明显优于传统非稳健中继波束成形算法。(4)本文研究了大规模天线阵列的宽带非理想特性,着重分析了大规模天线下非零带宽对传统窄带多重信号分类(MUltiple SIgnal Classification,MUSIC)DOA估计算法的影响。结果表明,在大规模天线情况下,MUSIC算法性能存在一个临界带宽点。该结果有助于指导实际大规模天线阵列系统选择窄带信号处理方案还是宽带信号处理方案。针对大规模相控阵在宽带情况下出现的波束倾斜现象,分析了波束倾斜对无线通信系统的影响,主要表现为固定阵列增益损失与等效信道频率选择性。为消除等效频率选择性信道中由波束倾斜引起的深衰落点,本文基于恒包络零自相关(Constant Amplitude Zero Auto-Correlation,CAZAC)序列提出了一种修正模拟波束成形方法,并在毫米波单载波频域均衡(Single-Carrier FrequencyDomain Equalization,SC-FDE)系统中仿真验证了该方法的有效性。仿真结果表明,所提出的修正模拟波束形成方法能够减轻波束倾斜造成的性能损失。更多还原

【Abstract】 In 5G and future B5G/6G wireless networks,millimeter-wave(mm Wave)bands will be used to meet the capacity demand of future network due to its advantage of huge bandwidth.However,mm Wave signals experience higher propagation loss compared to sub-6GHz bands,which limit the network coverage.Thanks to the small wavelength of mm Wave signals,mm Wave systems can leverage massive multiple input multiple output(MIMO)antennas packed into a small form factor,which can provide significant beamforming gains to combat the severe propagation loss.Moreover,combined with the cooperative transmission technology,the network coverage can be further extended and the severe line-of-sight(Lo S)blockage can be avoided for mm Wave systems.However,limited by the current semiconductor industry development,the mm Wave massive MIMO system faces the challenges of high complexity,cost and power consumption.To address above challenges,the analog-digital hybrid array architecture and the low-resolution analog-to-digital converters(ADCs)/digital-to-analog converters(DACs)transceiver emerged.For different application scenarios,choosing suitable hardware architecture is a direct and effective solution to reduce system complexity.However,the low complexity hardware architecture and the wideband effect of large-scale antenna array lead to the degradation of system performance.Thus,robust communication signal processing algorithms are needed to combat the system performance loss caused by hardware impairments.Moreover,there are multiple nodes in relay systems,which make the optimal design of the system more complicated.To this end,this dissertation mainly studies the low-complexity hardware architecture for mm Wave massive MIMO systems,and proposes effective high-precision relay channel estimation algorithms and robust beamforming algorithms to improve the system performance.The main contents and contributions of this dissertation are summarized as follows.Firstly,this dissertation discusses an overlapped subarray between traditional fullyconnected and sub-connected structures and determine how many antennas should be connected for each radio frequency(RF)chain in hybrid antenna array.Moreover,for the mm Wave downlink multi-user MIMO communication,this dissertation designs a twostage hybrid beamforming algorithm based on the overlapped subarray structure.Simulation results indicate that there exists the most cost-effective overlapped subarray spacing associated with the required number of phase shifters for the sparse mm Wave channel.Our analysis provides a guideline for the design of hybrid antenna array architecture.Secondly,this dissertation proposes a series of high-precision two-timescale channel estimation algorithms by exploring the two-timescale characteristics of dual-hop channel in practical MIMO relay systems.Specifically,in mobile cellular networks,the location of the macro-cell base station(BS)and the relay station(RS)are generally fixed,whereas the user equipments(UE)are in moving state.Thus,the dual-hop MIMO channel follows the two-timescale characteristics,i.e.,the UEs-RS channel varies in a short-timescale manner,whereas the RS-BS channel varies in a long-timescale manner.To this end,this dissertation formulates the two-timescale channel estimation problem as an approximate maximum likelihood(AML)problem with the one-stage training(OST)scheme.Then,a batch(AML-Batch)algorithm is designed to solve the resultant optimization problem.Moreover,to overcome the drawbacks of batch algorithm,a low-complexity online(AML-Online)algorithm is proposed based on the two-stage stochastic optimization framework.Simulation results show that the performance of the proposed two-timescale estimation algorithms is better than that of the traditional single-timescale estimation algorithms.Furthermore,considering mm Wave cooperative systems with the intelligent reflection surface(IRS),this dissertation proposes a two-timescale compressed channel estimation algorithm by exploring the structure sparsity characteristics of the cascaded channel,which can reduce the pilot overhead and improve the estimation accuracy.Thirdly,this dissertation studies the robust beamforming problem for massive MIMO multi-pair two-way relay systems with low-resolution ADCs/DACs to reduce the performance loss caused by low-resolution ADCs/DACs and obtain satisfactory performance in massive user access scenarios,Based on the separated minimum mean square error(MMSE)scheme,this dissertation proposes a robust MMSE(R-MMSE)beamforming algorithm.Simulation results show that the proposed R-MMSE algorithm can reduce the performance loss caused by low-resolution ADCs/DACs,and the performance of proposed R-MMSE algorithm is better than that of the traditional nonrobust algorithms.Finally,this dissertation studies the array wideband effect on wideband mm Wave ultra-massive antenna array systems.When the number of antennas and the signal bandwidth are very large,the narrowband array signal processing model is not suitable.Thus,the signal must be treated as a wideband signal.this dissertation analyzes the influence of non-zero bandwidth on narrowband multiple signal classification(MUSIC)algorithm in ultra-massive antenna array systems,and conclude that there exists a critical bandwidth point.Then,we study the wideband effect on wideband mm Wave ultramassive phased array.In the phased array,the wideband effect is also called as beam squint effect.With the standard analog beamforming method,we view the analog beamforming and the physical channel as a spatial equivalent channel.It can be observed that the beam squint causes obvious beamforming gain loss and the spatial equivalent channel becomes a frequency-selective channel with multiple deep fading points.To eliminate the deep fading points,an advanced analog beamforming method is proposed based on constant envelope zero auto-correlation(CAZAC)sequence.Finally,the biterror-rate(BER)performance of mm Wave single-carrier frequency-domain equalization(SC-FDE)system is simulated to verify the effectiveness of the proposed method.更多还原