Collaborative Research: Wideband Multi-Beam Antenna Arrays: Low-Complexity Algorithms and Analog-CMOS Implementations

联邦通信委员会认识到无线行业需要探索28-95千兆赫毫米波（mm-wave）频段，其中有更宽的带宽可用，未来的分配可能达到100千兆赫以上。毫米波带宽的爆炸式增长开启了5G无线系统的应用，包括通信、定位、成像和雷达。该项目解决了在毫米波频率上产生多个平行无线电“波束”的基本科学和工程挑战。无线电波束是指为无线通信和遥感建立点对点接触的定向信道。形成大量具有高带宽的这种无线电波束的能力将极大地改善下一代无线系统的性能。例如，多波束对于实现无线应用爆炸性增长所需的容量、数据速率和地理渗透的数量级增长至关重要。此外，它们对发射器和接收器都很重要。该项目将利用空间傅里叶变换和薄光学透镜之间的类比来获得多个宽带光束。与文献中基于透镜天线的方法不同，该项目将使用平面孔径天线与模拟集成电路相结合，产生许多受功率和尺寸限制的宽带毫米波波束。这种高度集成的方法对于5G智能设备、物联网、移动机器人、无人机等移动应用以及其他以毫米波为重点的新兴应用具有吸引力。除科学研究外，该项目还将确保少数民族学生和女学生在数学、通信以及微波电路和系统方面的职业生涯中得到指导。开发阵列信号处理、微波集成电路设计、超高速模拟信号处理等教学材料。首席研究员Madanayake和Mandal将组织一次小型会议，以加强俄亥俄州东北部附近大学的微波和毫米波研究活动。PI Madanayake将与共同PI合作，指导代表性不足的学生在科学、技术、工程和数学（STEM）领域的职业发展。提案团队在促进跨越电气工程和数学领域的STEM主题方面具有独特的地位。该项目将导致教育更广泛的社区跨学科合作的重要性。此外，该团队将努力展示学习更深层次的数学主题对于在技术和工程职业生涯中取得成功的重要性。多波束阵列接收机由于其信号流图的潜在复杂性，在集成电路形式下很难实现。本文提出了基于i)结构化复矩阵稀疏分解理论和ii)近似变换理论的数学方法来解决这个问题。所得到的矩阵是用模拟ic在多ghz带宽下实现的。其中一个智力贡献是基于延迟范德蒙德矩阵（DVM）稀疏分解的高效宽带波束形成器的发展。该DVM算法解决了长期存在的“波束斜视”问题，即波束方向随着输入频率的变化而改变，无法实现真正的宽带操作。另一个是推导具有近似离散傅里叶变换（DFT）的特定性质的变换矩阵。这种近似变换不受精确DFT的已知计算复杂度限制，并且基于近似DFT的多波束形成器实际上可以使用电流模模拟ic有效地实现。最后，探讨精密电路设计、数字校准、内置自检等方法，以模拟IC形式有效实现所提出的多波束形成网络。

The Federal Communications Commission recognizes the need for the wireless industry to explore the 28-95 GHz millimeter-wave (mm-wave) bands where wider bandwidth is available, and future allocations may reach above 100 GHz. This explosion of mm-wave bandwidth opens up applications in 5G wireless systems spanning communications, localization, imaging, and radar. This project addresses fundamental scientific and engineering challenges in generating multiple parallel radio "beams" at mm-wave frequencies. A radio beam refers to a directional channel that establishes point-to-point contact for wireless communications and remote sensing. The ability to form a large number of such radio beams with high bandwidths will tremendously improve the performance for next-generation wireless systems. For example, multiple beams are essential for achieving the orders-of-magnitude increases in capacity, data rate, and geographical penetration required by the explosive growth in wireless applications. Moreover, they are important for both transmitters and receivers. The project will draw on an analogy between the spatial Fourier transform and a thin optical lens to obtain multiple wideband beams. Unlike lens-antenna-based approaches in the literature, this project will use a planar aperture antenna in conjunction with analog integrated circuits to generate many wideband mm-wave beams subject to power and size constraints. The proposed highly integrated approach is attractive for mobile applications including 5G smart devices, the internet of things, mobile robotics, and unmanned aerial vehicles, and other emerging applications focused on mm-waves. In addition to scientific research, the project will ensure that both minority students and female students will be mentored towards careers in mathematics, communications, as well as microwave circuits and systems. Educational materials will be developed for teaching array signal processing, microwave integrated circuit (IC) design, and ultra-high-speed analog signal processing. Principal Investigators (PIs) Madanayake and Mandal will organize a mini-conference to enhance microwave and mm-wave research activities at nearby universities in northeast Ohio. PI Madanayake will collaborate with co-PIs towards mentoring underrepresented students towards careers in Science, Technology, Engineering, and Math (STEM). The proposal team is uniquely placed to promote STEM topics spanning both electrical engineering and mathematics domains. The project will lead to education of the wider community on the importance of cross-disciplinary collaboration. Further, the team will strive to show the importance of learning deeper math topics towards success in technology and engineering careers.A multi-beam array receiver is deeply difficult to realize in IC form due to the underlying complexity of its signal flow graph. In this work, mathematical methods based on the theories of i) sparse factorization of structured complex matrices, and ii) approximate transforms are proposed to solve this problem. The resulting matrices are realized with multi-GHz bandwidths using analog ICs. One of the intellectual contributions is the development of efficient wideband beamformers based on sparse factorizations of delay Vandermonde matrices (DVM). This DVM algorithm solves the longstanding "beam squint" problem, i.e., the fact that the beam direction changes with input frequency, making true wideband operation impossible. Another is the derivation of transform matrices with specified properties that approximate the discrete Fourier transform (DFT). Such approximate transforms are not subjected to the known computational complexity bounds of the exact DFT, and approximate-DFT-based multi-beamformers can in fact be efficiently implemented using current-mode analog ICs. Finally, precision circuit design, digital calibration, built-in self-test, and other methods will be explored for efficiently realizing the proposed multi-beamforming networks in analog IC form.