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 GHz毫米波(mm波)频段，其中有更宽的带宽可用，并且未来的分配可能达到100 GHz以上。毫米波带宽的爆炸式增长开启了5G无线系统的应用，涵盖通信、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联系的定向信道。能够形成大量高带宽的这种无线电波束将极大地提高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率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，该项目将结合模拟集成电路使用平面孔径天线来产生许多受功率和尺寸限制的宽带毫米波波束。拟议的高度集成方法对移动应用具有吸引力，包括5G智能设备、物联网、移动机器人和无人机，以及其他专注于毫米波的新兴应用。除了科学研究，该项目还将确保少数民族学生和女性学生都能在数学、通信以及微波电路和系统等领域获得职业指导。将开发用于教授阵列信号处理、微波集成电路(IC)设计和超高速模拟信号处理的教材。首席研究员Madanayake和Mandal将在俄亥俄州东北部附近的大学组织一次小型会议，以加强微波和毫米波研究活动。Pi Madanayake将与合作PI合作，指导代表人数不足的学生走向科学、技术、工程和数学(STEM)的职业生涯。提案团队在推动STEM主题跨越电气工程和数学领域方面具有得天独厚的优势。该项目将对更广泛的社区进行教育，使他们了解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。由于其信号流图的潜在复杂性，多波束阵列接收器很难以IC形式实现。在这项工作中，提出了基于i)结构复矩阵的稀疏分解和ii)近似变换理论的数学方法来解决这个问题。所得到的矩阵是使用模拟IC以多GHz带宽实现的。基于延迟Vandermonde矩阵(DVM)稀疏分解的高效宽带波束形成器的发展是其智力贡献之一。这种DVM算法解决了长期存在的“波束斜视”问题，即波束方向随输入频率变化的事实，使得真正的宽带运算变得不可能。另一种是具有近似离散傅立叶变换(DFT)的特定性质的变换矩阵的推导。这种近似变换不受精确DFT的已知计算复杂度的限制，基于近似DFT的多波束形成器实际上可以使用电流模式模拟IC来高效地实现。最后，我们将探索精密电路设计、数字校准、内建自测试等方法，以有效地实现所提出的模拟IC形式的多波束形成网络。

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.