SWIFT: LARGE: Adaptive Radio Frequency Interference Cancellation for Radio Science Observatories

SWIFT：LARGE：无线电科学天文台的自适应射频干扰消除

• 批准号: 2029670
• 负责人: Frank Lind
• 金额: $ 125.97万
• 依托单位: Massachusetts Institute of Technology
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2021
• 资助国家: 美国
• 起止时间: 2021-01-01 至 2024-12-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2029670&HistoricalAwards=false
• 关键词: SWIFT; LARGE; Adaptive; Radio; Frequency

Scientific radio observatories, including radar systems and radio telescopes, require access to the radio spectrum for experiments. In some cases the experiments many be conducted in frequencies that already overlap communications systems or which are subject to unintentional radio frequency interference (RFI). By exploring techniques for mitigating the interference we will help to preserve the integrity of scientific measurements and improve the ability of radio observatories to operate in a future environment of radio spectrum sharing. The techniques developed will focus on automated methods to identify and mitigate radio interference. The approaches taken will include advanced mathematical methods and artificial intelligence techniques. These methods will be highly efficient compared to existing approaches and usable immediately to help improve the quality of radio observatory data. The underlying techniques and algorithms are likely to be highly useful beyond the immediate context of this proposal. They will be made available as open source software in the Julia programming language to enable widespread use by the scientific community. Scientific radio observatories including radar systems and radio telescopes require access to the radio spectrum for experiments. In some cases the experiments are limited to bands which can be protected but many are conducted in frequencies that already overlap existing communications systems or are subject to unintentional radio frequency interference (RFI). We aim to explore techniques for spectral coexistence in the context of experimental radio science for Geospace and Astronomy applications. Our approach is to explore the development and application of algorithms and mathematical techniques for adaptive cancellation of RFI for systems at MIT Haystack Observatory. Modern radio observatories must co-exist with a vast increase in both licensed and unintentional sources of RFI. The experiments at these observatories often require the measurement of signals that are significantly below the background thermal noise floor. A core goal of the effort is to develop algorithms and methodologies that will enable interference-immune radio science systems. Starting from existing approaches to space time adaptive filtering (STAP) and a sidelobe cancelling architecture we will implement a sparse reference sensor network and explore: (1) automated methods for identification and classification of RFI focused, (2) random matrix, sketch, and neural network techniques to accelerate the algorithms, and (3) the application of sparse software radio networks to provide simultaneous RFI mitigation for multiple radio observatory sensors. Our proposed work will enable advanced algorithms for adaptive cancellation of RFI in the context of a large scale radio facility. The approaches should be generally applicable to a wider range of radio facilities and experiments. In many cases the adaptive cancellation techniques may enable observations in spectrum that might otherwise not be usable. Mathematical approaches to the efficient application of these algorithms and their scaling using high performance computing will be demonstrated. Both the underlying techniques and the specific algorithms are likely to be highly useful beyond the immediate context of this proposal. Project results will be disseminated in venues, spanning communities from signal processing, computer science, mathematics, and Radio Science (i.e. URSI). The scope was slightly reduced after review to focus on four areas of the proposal associated with the multi-channel sidelobe cancellation architecture instead of five, with the specific reduction being determined as the work progresses. While the major project goals remain unchanged, as part of the scope reduction, the team will utilize smaller datasets to reduce computational costs.This 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.

包括雷达系统和射电望远镜在内的科学射电天文台需要利用无线电频谱进行实验。在某些情况下，实验可能在已经重叠通信系统的频率上进行，或者在受到无意射频干扰（RFI）的频率上进行。通过探索减轻干扰的技术，我们将有助于保持科学测量的完整性，并提高无线电观测站在未来无线电频谱共享环境中的运行能力。开发的技术将侧重于识别和减轻无线电干扰的自动化方法。所采取的方法将包括先进的数学方法和人工智能技术。与现有方法相比，这些方法将非常有效，并可立即用于帮助提高射电天文台数据的质量。潜在的技术和算法可能非常有用，超出了本提案的直接背景。它们将以Julia编程语言的开源软件形式提供，以使科学界能够广泛使用。包括雷达系统和射电望远镜在内的科学射电天文台需要利用无线电频谱进行实验。在某些情况下，实验仅限于可以保护的频带，但许多实验是在已经与现有通信系统重叠的频率上进行的，或者受到无意的射频干扰（RFI）。我们的目标是探索在地球空间和天文学应用的实验无线电科学背景下的光谱共存技术。我们的方法是探索麻省理工学院干草堆天文台系统自适应取消RFI的算法和数学技术的发展和应用。现代射电天文台必须与许可的和无意的RFI源的大量增加共存。这些天文台的实验通常需要测量明显低于背景热噪声底的信号。这项工作的核心目标是开发算法和方法，使抗干扰无线电科学系统成为可能。从现有的时空自适应滤波（STAP）和旁瓣消除体系结构的方法出发，我们将实现一个稀疏参考传感器网络，并探索：(1)识别和分类RFI聚焦的自动化方法，(2)随机矩阵、草图和神经网络技术来加速算法，以及(3)稀疏软件无线电网络的应用，为多个射电天文台传感器提供同时的RFI缓解。我们提出的工作将使先进的算法能够在大规模无线电设施的背景下自适应取消RFI。这些办法应普遍适用于范围更广的无线电设施和实验。在许多情况下，自适应消去技术可以使在光谱上的观测成为可能，否则这些观测是不可用的。数学方法来有效地应用这些算法和它们的缩放使用高性能计算将被证明。基础技术和具体算法都可能在本提案的直接上下文之外非常有用。项目结果将在各个场所传播，涵盖信号处理、计算机科学、数学和无线电科学（即URSI）等社区。审查后，范围略有减少，将重点放在与多通道旁瓣对消架构相关的提案的四个领域，而不是五个领域，具体减少的范围将随着工作的进展而确定。虽然主要的项目目标保持不变，但作为范围缩小的一部分，团队将利用更小的数据集来减少计算成本。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Fifty Three Matrix Factorizations: A Systematic Approach

五十三次矩阵分解：系统方法

• DOI: 10.1137/21m1416035
• 发表时间: 2023
• 期刊: SIAM Journal on Matrix Analysis and Applications
• 影响因子: 1.5
• 作者: Edelman, Alan;Jeong, Sungwoo
• 通讯作者: Jeong, Sungwoo

On the structure of the solutions to the matrix equation G⁎JG = J

矩阵方程G−JG−=−J的解的结构

• DOI: 10.1016/j.laa.2022.10.007
• 发表时间: 2023
• 期刊: Linear Algebra and its Applications
• 影响因子: 1.1
• 作者: Edelman, Alan;Jeong, Sungwoo
• 通讯作者: Jeong, Sungwoo

Symbolic-numeric integration of univariate expressions based on sparse regression

基于稀疏回归的单变量表达式的符号数值积分

• DOI: 10.1145/3572867.3572882
• 发表时间: 2022
• 期刊: ACM Communications in Computer Algebra
• 影响因子: 0.1
• 作者: Iravanian, Shahriar;Martensen, Carl Julius;Cheli, Alessandro;Gowda, Shashi;Jain, Anand;Ma, Yingbo;Rackauckas, Chris
• 通讯作者: Rackauckas, Chris