Skip to main content
QUICK REVIEW

[论文解读] Hybrid Analog-Digital Transceiver Designs for Cognitive Large-Scale Antenna Array Systems

Christos G. Tsinos, Sina Maleki|arXiv (Cornell University)|Dec 9, 2016
Advanced MIMO Systems Optimization参考文献 41被引用 7
一句话总结

本文提出了两种用于认知毫米波大规模MIMO系统的混合模拟-数字预编码设计,可在保护现有用户的同时实现频谱共享,并最小化干扰。通过采用基于ADMM的优化框架,该方法在显著降低硬件复杂度和功耗的同时,实现了接近全数字波束成形的频谱效率,使其在实际毫米波部署中具备可行性。

ABSTRACT

Milimeter wave (mmWave) band mobile communications can be a solution to the continuously increasing traffic demand in modern wireless systems. Even though mmWave bands are scarcely occupied, the design of a prospect transceiver should guarantee the efficient coexistence with the incumbent services in these bands. To that end, in this paper, two underlay cognitive transceiver designs are proposed that enable the mmWave spectrum access while controlling the interference to the incumbent users. MmWave systems usually require large antenna arrays to achieve satisfactory performance and thus, they cannot support fully digital transceiver designs due to high demands in hardware complexity and power consumption. Thus, in order to develop efficient solutions, the proposed approaches are based on a hybrid analog-digital pre-coding architecture. In such hybrid designs, the overall beamformer can be factorized in a low dimensional digital counterpart applied in the baseband and in an analog one applied in the RF domain. The first cognitive solution developed in this paper designs the cognitive hybrid pre-coder by maximizing the mutual information between its two ends subject to interference, power and hardware constraints related to the analog counterpart. The second solution aims at reduced complexity requirements and thus derives the hybrid pre-coder by minimizing the Frobenious norm of its difference to the optimal digital only one. A novel solution for the post-coder at the cognitive receiver part is further proposed here based on a hardware constrained Minimum Mean Square Error criterion. Simulations show that the performance of both the proposed hybrid approaches is very close to the one of the fully digital solution for typical wireless environments.

研究动机与目标

  • 解决在认知无线电系统中实现毫米波频段频谱接入的同时保护现有用户的技术挑战。
  • 克服大规模毫米波MIMO系统中全数字预编码带来的高硬件复杂度和高功耗问题。
  • 设计结合低维数字处理与模拟移相器网络的混合预编码架构,以降低功耗和成本。
  • 在最大化认知毫米波系统频谱效率的同时,确保对主用户造成的干扰低于预设阈值。
  • 在接收端设计一种硬件受限的最小均方误差(MMSE)后均衡器,以在实际约束条件下提升性能。

提出的方法

  • 建立混合预编码问题,以在干扰、功率和模拟硬件约束下最大化收发端之间的互信息。
  • 采用交替方向乘子法(ADMM)通过交替更新数字预编码器、模拟波束成形器和对偶变量,高效求解非凸优化问题。
  • 提出第二种低复杂度设计,其目标是最小化混合预编码器与最优全数字预编码器之间差异的Frobenius范数。
  • 在接收端提出一种硬件受限的MMSE后均衡器,以缓解残留干扰并提升在实际射频链路限制下的信号检测性能。
  • 对模拟波束成形器施加常模约束,以反映实际移相器硬件的限制。
  • 应用ADMM迭代优化预编码器和合并器,同时通过算法收敛特性理论分析确保收敛性。

实验结果

研究问题

  • RQ1混合模拟-数字预编码设计是否能在尊重对现有用户干扰约束的前提下,实现在认知毫米波大规模MIMO系统中接近最优的频谱效率?
  • RQ2所提出的混合预编码方案在频谱效率和干扰调控方面与全数字波束成形相比表现如何?
  • RQ3在毫米波认知无线电系统中,混合预编码架构在性能与硬件复杂度之间存在何种权衡?
  • RQ4基于ADMM的优化框架是否能在实际硬件约束下保证收敛性,并获得稳定且接近最优的解?
  • RQ5所提出的硬件受限MMSE后均衡器在射频链路和移相器约束受限条件下,对提升接收机性能的有效性如何?

主要发现

  • 所提出的混合预编码设计即使在射频链路数量减少的情况下,也能实现与全数字预编码器非常接近的频谱效率。
  • 基于ADMM的优化算法收敛至满足Karush-Kuhn-Tucker(KKT)条件的解,表明其具有局部最优性。
  • 在典型的毫米波传播环境中,混合系统与全数字系统之间的性能差距极小,验证了混合设计的实用性。
  • 基于Frobenius范数最小化的低复杂度设计在显著降低计算负载的同时,实现了接近最优的性能。
  • 硬件受限的MMSE后均衡器能有效抑制干扰,并在实际射频链路限制下提升信号检测精度。
  • 理论分析证实,ADMM序列收敛至满足KKT条件的极限点,确保了在给定约束下解的最优性。

更好的研究,从现在开始

从阅读论文到最终审阅,大幅缩短您的研究时间。

无需绑定信用卡

本解读由 AI 生成,并经人工编辑审核。