SWIFT-SAT: Efficient and On-Demand Spectrum Coexistence for Satellite-Terrestrial Systems

在新兴应用的推动下，对无线通信的需求不断上升，导致无线电频谱拥挤。本课题设计了卫星与地面系统频谱共存的有效方案。目前使用的一种频谱共享方法被称为频谱接入系统（SAS），这是一种集中的数据库和控制系统，可以实时为用户分配信道，从而保护其他用户免受干扰。现有的SAS解决方案支持在每个设备都进行传输和接收的情况下实现地面与地面共享。本项目开发的新方案将SAS概念扩展到管理卫星-地面频谱共享。该项目解决的主要挑战包括：(a)地面和轨道上的一些设备是被动接收器，无法通过频谱传感检测到，以及(b) SAS必须管理多个发射器的聚合干扰，并感知动态变化的接收器波束。该项目的研究结果将为政府决策者开放中频段和高频段以供频谱共享和商业使用的持续努力提供有价值的见解。该项目有三个主要的研究重点。第一部分探讨了卫星-地面SAS的架构选择。该体系结构的关键考虑因素包括：关于系统、位置和梁的有效数据收集；集束干扰的高效计算；共存解决方案的有效计算，这可能涉及由受影响用户运行的可扩展分散算法；并尽量减少对现有卫星系统协议的更改。第二个重点是发展有效的基于sas的方法来实现卫星系统的干扰保护，包括信道估计和综合干扰计算。考虑的因素包括定向天线、先进的干扰管理方案（如MIMO和连续干扰消除）、地形阻塞、空闲时间、部分频谱利用和基于风险的干扰保护。第三个推力设计了一个多时间尺度共存方案：针对具有固定波束或地面系统的卫星的预计算解决方案，以及针对具有不可预测波束和移动地面系统的卫星的实时计算解决方案。将初始非线性优化问题线性化，得到混合整数线性规划（MILP）公式。开发了多种算法来解决MILP问题：使用子问题分区来实现gpu上的实时优化，使用强化学习来开发低复杂度算法，并使用联邦学习来实现跨地面用户网络的分散解决方案。研究的评估包括候选架构的技术和政策评估，通道预测的验证以及共存算法的软件原型。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。

The escalating demand for wireless communications, fueled by emerging applications, has led to radio spectrum congestion. This project designs efficient schemes for spectrum coexistence between satellite and terrestrial systems. One spectrum sharing approach used today is called a Spectrum Access System (SAS), a centralized database and control system that assigns channels to users in real-time such that other users are protected from interference. Existing SAS solutions support terrestrial-terrestrial sharing in situations where every device transmits as well as receives. The new schemes developed in this project extend the SAS concept to manage satellite-terrestrial spectrum sharing. Key challenges addressed in the project include (a) some devices on the ground and in orbit are passive receivers that cannot be detected by spectrum sensing, and (b) the SAS must manage aggregate interference of multiple transmitters with awareness of dynamically changing receiver beams. The findings from this project will provide valuable insights for ongoing efforts by government policy-makers to open mid-bands and high-bands for spectrum sharing and commercial use.The project has three main research thrusts. The first thrust explores architectural options for the satellite-terrestrial SAS. Key considerations for the architecture include: effective data collection regarding systems, locations, and beams; efficient calculation of aggregate interference; efficient calculation of coexistence solutions, which may involve scalable decentralized algorithms run by affected users; and minimization of changes to existing satellite system protocols. The second thrust develops effective SAS-based methods to achieve interference protection for satellite systems, including channel estimation and aggregate interference computation. Factors considered include directional antennas, advanced interference management schemes such as MIMO and successive interference cancellation, terrain blockage, idle periods, partial spectrum utilization, and risk-based interference protection. The third thrust designs a multi-time scale coexistence scheme: precalculated solutions for satellites with fixed beams or ground systems, and real-time calculated solutions for satellites with unpredictable beams and mobile ground systems. The initially nonlinear optimization problem is linearized to obtain Mixed Integer Linear Program (MILP) formulations. Multiple algorithms to solve the MILP problems are developed: use subproblem partitioning to enable real time optimization on GPUs, use reinforcement learning to develop low-complexity algorithms, and use federated learning to enable decentralized solution across a network of terrestrial users. Evaluation of the research includes technical and policy assessment of candidate architectures, verification of channel predictions, and software prototypes of coexistence algorithms.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.