【作者】 郭宁；

【导师】 黑永强；

【作者基本信息】 西安电子科技大学 ， 通信与信息系统， 2019， 硕士

【摘要】 毫米波MIMO通信系统提供了大量可用的频谱资源,且支持吉比特级的高速传输,因此得到了广泛的关注。针对毫米波MIMO通信中路径损耗严重的问题,通常采用波束赋形技术获取阵列增益以对抗路径损耗。然而对于毫米波MIMO通信系统而言,全数字波束赋形技术需要数量巨大射频链路,其信号处理复杂性和成本开销无疑对实际系统带来了极大的挑战。在此背景下,混合波束赋形技术应运而生。混合波束赋形技术将波束赋形分成数字域和模拟域两部分进行,大大降低了所需的射频链路数及实现复杂度。本文针对毫米波MIMO通信系统,从不同的波束赋形架构入手,重点研究低功耗的混合波束赋形技术,主要工作概括如下:第一,研究了低功耗的开关-反向器混合波束赋形架构,分析了架构特点并重构了该结构下的问题模型。针对该架构仅能提供有限相位控制的特点,且该架构下已提出的基于交叉熵(CE)的混合波束赋形算法复杂度较高的问题,结合马尔科夫链蒙特卡洛(MCMC)方法中的Gibbs采样思想,提出了基于Gibbs采样的混合波束赋形方案。该算法从初始样本出发,逐位置更新各个坐标上的元素,通过有限次迭代获得模拟波束赋形矩阵。仿真结果表明,所提算法在该架构下能够获得更高的频谱效率,且相较于CE算法复杂度更低。第二,借鉴相控阵中相位调制阵列(PMA)的原理,提出了一种新的基于PMA的混合波束赋形架构,通过引入时间自由度,利用低功耗的射频开关与延迟线代替模拟移相器实现模拟部分的波束控制。首先理论分析了该架构的性能,随后通过仿真证明了此架构低功耗、高精度相位控制以及等效幅度控制的优势。针对其打破了移相器恒模限制的特点,搭建了该结构下的混合波束赋形问题模型。针对该结构能在模拟部分实现等效幅相控制的特性,首先对相位提取ZF(PZF)波束赋形算法进行了改进,此外,结合粒子群(PSO)优化算法,提出了基于PSO的混合波束赋形算法,将模拟波束赋形矩阵的中间变量看做粒子,通过更新粒子群中粒子的速度和位置优化相位与时间变量,以此获得期望的波束赋形矩阵。仿真结果证明,所提混合波束赋形方案相对于传统混合波束赋形方案功耗更低,能量效率更高,且能够达到近似最优的频谱效率。更多还原

【Abstract】 Millimeter wave(mmWave)MIMO communication has a promising prospect due to its merit of rich spectrum resource and multigigabit-per-second data rate.In order to solve the problem of high path loss in mmWave MIMO communication systems,the beamforming is usually used to obtain the array gain to overcome the path loss.However,for the mmWave MIMO communication systems,the full digital beamforming requires a large number of RF links and the complexity and cost of signal processing undoubtedly bring great challenges to the actual systems.Therefore,hybrid beamforming technology emerges at the right moment.The beamforming is divided into the digital domain and the analog domain in the hybrid beamforming,which greatly reduces the number of the required RF links and the complexity of implementation.Aiming at the millimeter-wave MIMO communication system,this paper focuses on the low-power hybrid beamforming from the perspective of different beamforming architectures.The main work is summarized as follows: Firstly,the characteristics of the Switches and Inverters-based architecture are analyzed,and the problem model under the structure is reconstructed.Aiming at the problem that the architecture can only provide limited phase control and the complexity of the cross entropy(CE)-based hybrid beamforming algorithm is high,combining with Gibbs sampling idea in markov chain monte carlo(MCMC)method,the Gibbs sampling-based hybrid beamforming algorithm is proposed.Starting from the initial samples,the elements on each coordinate are updated position by position in the algorithm,then the beamforming matrix can be obtained by finite iterations.Simulation results show that the proposed algorithm can achieve reasonable spectral efficiency and lower algorithm complexity.Secondly,combined with the phase modulation array(PMA)in the phased array,a PMA-based hybrid beamforming architecture is proposed.By introducing time degrees of freedom,low-power RF switches and delay lines are used to replace the analog phase shifter to realize the beam control in the analog part.The characteristics of this architecture are studied then the advantages of low power consumption,high precision phase control and equivalent amplitude control are proved by theoretical analysis.According to its characteristic of breaking the constant modulus constraint of the phase shifter,the problem model under this structure is reconstructed.First,due to the phase-amplitude control in the analog part,the phase extraction ZF(PZF)beamforming algorithm is improved.In addition,combined with the idea of the particle swarm optimization(PSO)algorithm,the PSO-based hybrid beamforming algorithm is proposed.The intermediate variables of the anolog beamforming matrix are regarded as particles,and the anolog beamforming matrix is obtained by updating the velocity and position of the particles in the particle swarm to optimize the required phase and time.Simulation results show that the proposed scheme has lower power consumption,higher energy efficiency(EE)and approximately optimal spectrum efficiency(SE)compared with the traditional hybrid beamforming schemes.更多还原