Collaborative Research: SpecEES: Towards Energy and Spectrally Efficient Millimeter Wave MIMO Platforms - A Unified System, Circuits, and Machine Learning Framework

无线网络自诞生以来一直发挥着变革性的社会作用。下一代无线网络的发展是美国和世界其他国家的国家优先事项。截至2019年，5G无线网络正处于部署的早期阶段，目标是能够支持数十亿设备上快速增长的低延迟移动数据流量，同时降低整体网络能耗和成本。最先进形式的 5G 网络的部署预计需要数年时间，一直持续到 2020 年代。然而，消费者对无线网络容量的需求预计将继续有增无减，特别是随着自动运输和交付网络等新应用场景的成熟。考虑到此类需求，该项目旨在通过采用涵盖电路、系统和人工智能的统一方法来研究“超越 5G”网络的硬件和物理层需求。所提出的理论、算法和硬件实现预计将在许多领域产生影响，包括向行业的技术转让、本科生和研究生课程材料的开发、研究生培训、本科生研究经验和通过无线测试台开发进行的社区推广，以及所有模拟框架和机器学习数据集的公开发布。 该项目的研究目标是开发一套用于毫米波MIMO系统的分析和设计工具，包括特定的电路感知信号处理技术和新颖的算法感知电路设计。将为解决毫米波 MIMO 系统的关键挑战做出基础算法贡献，例如提高高度移动应用和密集毫米波部署中的频谱效率和能源效率。基础电路贡献将包括设计节能 MIMO 发射机、设计节能和面积高效的射频预编码器和组合器以及设计支持机器学习算法的平台的解决方案。该项目有几个相互关联的主旨：（1）研究混合架构的联合系统/电路分析和设计方法； (2) 开发新颖的电路（包括高效发射机和双向信号路径），以在混合 MIMO 架构中实现高能效、可重构性和并发多频段操作。 (3) 采用机器学习工具来设计电路和部署感知的波束赋形码本，并利用机器学习技术来设计毫米波干扰感知波束赋形； (5) 使用适当的传感器和执行器集成到 MIMO 平台中，以实现上述机器学习技术的硬件实现该奖项反映了 NSF 的法定使命，并通过使用基金会的智力优点和更广泛的影响审查标准进行评估，被认为值得支持。

Wireless networks have played a transformative societal role since their inception. The development of the next generation of wireless networks is a national priority for the United States and other countries around the world. As of 2019, 5G wireless networks are in early stages of deployment, and have the goals of being able to support rapidly increasing mobile data traffic with low latency across billions of devices, while reducing overall network energy consumption and cost. The deployment of 5G networks in their most advanced form is expected to take several years lasting well into the 2020's. However, consumer demands that drive wireless network capacity is projected to continue unabated, especially as new application scenarios such as autonomous transportation and delivery networks mature. In anticipation of such needs, this project seeks to investigate hardware and physical layer needs of "beyond 5G" networks by taking a unified approach that encompasses circuits, systems and artificial intelligence. The proposed theories, algorithms, and hardware implementation are expected to have impacts in a number of areas that include technology transfer to industry, development of undergraduate and graduate course materials, graduate student training, undergraduate research experiences and community outreach via wireless testbed development, and public release of all simulation frameworks and machine learning datasets. The research goal of this project is to develop a set of analysis and design tools for mm-wave MIMO systems including specific circuits-aware signal processing techniques, and novel algorithms-aware circuit designs. Fundamental algorithmic contributions will be made to solve key mm-wave MIMO system challenges such as enhancing the spectral efficiency and energy efficiency in highly-mobile applications and dense mm-wave deployments. Fundamental circuit contributions will include solutions to designing energy-efficient MIMO transmitters, designing energy- and area-efficient RF precoders and combiners, and designing platforms to support machine learning algorithms. The project has several inter-related thrusts: (1) Investigate joint system/circuit analysis and design approaches for hybrid architectures; (2) Develop novel circuits (including high-efficiency transmitters and bi-directional signal paths) to enable high energy-efficiency, reconfigurability and concurrent multi-band operation in hybrid MIMO architectures (3) Adopt machine learning tools to design circuits- and deployment-aware beamforming codebooks, and leverage machine learning techniques to design mm-wave interference-aware beamforming; (5) Integrate into MIMO platforms with appropriate sensors and actuators to enable hardware implementation of the aforementioned machine learning techniquesThis award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.