登录 EN

添加临时用户

面向移动覆盖的太赫兹大规模多入多出传输技术研究

Terahertz Massive Multiple--Input Multiple-Output Technology for Mobile Coverage

作者:谭竞搏
  • 学号
    2017******
  • 学位
    博士
  • 电子邮箱
    188******com
  • 答辩日期
    2022.05.19
  • 导师
    戴凌龙
  • 学科名
    信息与通信工程
  • 页码
    108
  • 保密级别
    公开
  • 培养单位
    023 电子系
  • 中文关键词
    太赫兹通信,大规模多入多出系统,预编码,信道估计,波束赋形
  • 英文关键词
    Terahertz communications, massive multiple--input multiple--output, precoding, channel estimation, beamforming

摘要

太赫兹通信可以提供数十倍于毫米波通信的带宽,从而支持超高速率传输,被认为是未来第六代移动通信(6th Generation Mobile Communication, 6G)网络的重要候选技术之一。但太赫兹频段信号面临严重的路径损耗,难以支持室外热点移动覆盖。因此,能够生成高增益窄波束的大规模多入多出(massive multiple-input multiple-output, massive MIMO)技术被认为具有应用于太赫兹通信的潜力。太赫兹大规模 MIMO 被认为是未来 6G 的关键技术之一。然而,太赫兹大规模 MIMO 因大带宽和高天线数面临严重的波束分裂问题。波束分裂导致太赫兹大规模 MIMO系统难以生成高增益波束,从而使得系统的移动覆盖能力显著降低。为此,本文围绕如何解决太赫兹大规模 MIMO 系统的波束分裂问题展开研究。首先,本文指出了太赫兹大规模 MIMO 系统中的波束分裂问题。在理论上分析了波束分裂的形成机理,证实了太赫兹大规模 MIMO 的大带宽和高天线数将导致严重的波束分裂。定义了衡量波束分裂严重程度的性能指标,即波束分裂比率。之后,针对波束分裂使得太赫兹大规模 MIMO 波束赋形损失大的问题,提出在传统模数混合预编码架构中引入少量延时模块的延时相移双重调控架构。该架构利用延时模块引入的频率选择性相移,基于证明的波束分裂补偿机理,可以消除波束分裂造成的性能损失。在此基础上,提出对应的预编码设计方法。提出的基于延时相移双重调控的预编码架构和对应的预编码方法能够实现准最优的阵列增益和可达速率性能。然后,针对波束分裂导致传统信道估计方法在太赫兹大规模 MIMO 系统中估计精度降低的问题,提出波束分裂模式匹配的宽带信道估计方法。在充分挖掘太赫兹大规模 MIMO 频域和角度域信道特征的基础上,定义了波束分裂模式,并证明了波束分裂模式与信道物理角度的一一对应关系。提出利用波束分裂模式估计路径角度和稀疏支撑集。所提方法较传统信道估计方法提升了信道估计的精度。最后,针对波束分裂导致传统波束追踪方法失效的问题,提出基于波束柔性分裂的波束追踪方法。利用延时相移双重调控架构,证明了能够灵活调控波束分裂程度的波束柔性分裂机理。基于该机理,提出利用同时生成的多个波束同时追踪多个物理角度。所提方法与传统波束追踪方法相比,能够大幅降低波束追踪开销,并提升波束追踪精度。

Terahertz (THz) communications can provide dozens of times the bandwidth of millimeter-wave communications which can support ultra-high-speed transmission, and is considered to be one of the important candidate technologies for the 6th Generation Mobile Communication (6G) network in the future. However, signals in the THz band suffer from severe path loss, making it difficult to support mobile coverage of outdoor hotspots. Therefore, massive multiple-input multiple-output (massive MIMO) technology capable of generating high-array-gain narrow beam is considered to have the potential to be applied to THz communication. Thus, THz Massive MIMO is considered to be one of thekey technologies for future 6G. However, THz massive MIMO suffers from severe beam split due to wide bandwidth and large number of antennas. Beam split makes it difficult for THz massive MIMO systems to generate high-gain beams, which significantly reduces the mobile coverage capability of THz massive MIMO systems. Therefore, this paper focuses on how to solve the beam split problem of THz massive MIMO systems.Firstly, this paper figures out the problem of beam split in THz massive MIMO systems. The formation mechanism of beam split is theoretically analyzed, and it is confirmed that the wide bandwidth and large number of antennas of THz massive MIMO system will lead to severe beam split. A performance indicator that measures the severity of beam split, the beam split ratio, is defined.Secondly, to solve the problem that beam split makes THz massive MIMO suffer from large array gain loss, a delay-phase precoding architecture is proposed, which introduces a small number of delay modules into the traditional analog-digital hybrid precoding architecture. The proposed architecture utilizes the frequency-dependent phase shift introduced by the delay module. Based on the proven beam split compensation mechanism, the performance loss caused by beam split can be efficiently eliminated. Then, a corresponding precoding design method is proposed. The proposed delay-phase precoding architecture and the corresponding precoding method can achieve near-optimal array gain and achievable rate performance. Then, to deal with the problem that beam split leads to the reduction of estimation accuracy of traditional channel estimation methods in THz massive MIMO systems, a beam split pattern based wideband channel estimation method is proposed. By fully exploiting the channel characteristics of THz massive MIMO in the frequency and angle domains, the beam split pattern is defined, and the one-to-one correspondence between the beam split pattern and the physical direction of the channel is proved. It is proposed to utilize beam split patterns to estimate channel path physical directions and corresponding sparse supports. Compared with the traditional channel estimation method, the proposed method improves the accuracy of channel estimation.Finally, a beam zooming based beam tracking method is proposed to cope with the problem that the traditional beam tracking method fails due to beam split. The beam zooming mechanism that can flexibly control the degree of beam split is proved by using the delay-phase precoding architecture. Based on this mechanism, it is proposed to use multiple beams generated simultaneously to track multiple channel physical directions simultaneously. Compared with the traditional beam tracking method, the proposed method can greatly reduce the beam tracking overhead and improve the beam tracking accuracy.