Quickest Spectrum Awareness under Correlated Spatio-temporal Variations

无线通信技术正在迅速超越其主要应用（例如，蜂窝通信），并作为影响不同技术、社会和经济领域的各种新兴技术的关键骨干，发挥着变革性作用。所有这些技术都是以增加对无线电频谱的需求为代价的，而无线电频谱是无线行业的关键商品。实现长期和可持续的无线基础设施需要提高频谱接入效率并允许实时和细粒度频谱管理的技术。频谱管理的有效性关键取决于形成跨时间和空间的频谱占用状态的敏捷和准确的态势感知。该研究项目的重点是开发一个全面的分析框架，以形成无线频谱机会时空变化的态势感知。该框架建立了与传感电路的双向交互。在前进方向上，它接收传感器收集的测量数据，以形成各种检测、推断和资源管理决策。相反，基于提供准确态势感知所需的数据，它会向传感器发送指令，告诉传感器如何继续从频谱的各个部分收集数据。这个框架形成的决策也可以用于可视化决策和警报。本项目的研究任务包括：开发一个综合框架，将创新频谱感知的几个关键组件连贯集成，形成敏捷的推理决策，实现动态环境下的快速计算和快速自适应，并确保环境不确定性下的决策鲁棒性。为了开发这样一个全面的框架，该项目有几个研究重点，每个研究重点都涉及框架的一个方面。在每个推力中，项目追求以下两个相互交织的研究目标：(1)理论基础：第一个目标是确定形成最快态势感知的基本限制，同时认识到由于传感成本、通信延迟容忍度、模型不确定性和频谱传感受益者的动态变化而产生的物理限制，其中包括在不同协议和基础设施下运行的各种自主无线应用。(2)算法设计：在发现基本限制的基础上，设计快速形成态势感知的数据分析工具和机制，实现实时、可扩展、决策保障、充分适应和响应宽带频谱占用状态的时空变化。该项目将产生一个决策理论框架，用于形成对宽带频谱不同段的频谱占用状态的快速可靠的态势感知。为了实现这一目标，该项目将解决长期存在的开放性问题，如快速检测网络数据的变化，控制传感中的开放性问题，以及由频谱传感物理约束引起的挑战性问题，如1)传感预算，2)模型不确定性下的决策，以及3)检测试图破坏传感数据和扭曲频谱传感决策的对手。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。

Wireless communication technology is rapidly transcending its primary applications (e.g., cellular communications) and plays a transformative role as the critical backbone in various emerging new technologies that affect different technological, social, and economic domains. All such technologies are viable at the expense of increasing demands for radio spectrum, which is the key commodity in the wireless industry. Enabling long-term and sustainable wireless infrastructures necessitates technologies that enhance spectrum access efficiency and allow for real-time and fine-granularity spectrum management. The effectiveness of spectrum management critically depends on forming agile and accurate situational awareness of spectrum occupancy states across time and space. This research project focuses on developing a comprehensive analytical framework for forming situational awareness of the spatio-temporal variations in wireless spectrum opportunities. This framework establishes bi-directional interactions with the sensing circuitry. In the forward direction, it receives measurements collected by the sensors in order to form various detection, inferential, and resource management decisions. In the reverse direction, based on the data deemed necessary for delivering accurate situational awareness, it sends instructions to the sensors on how to proceed with collecting data from various segments of the spectrum. The decisions formed by this framework can also be used for visualizing decisions and alarms. The research tasks of this project include: developing a comprehensive framework that coherently integrates several critical components for innovative spectrum sensing, forming agile inferential decisions, achieving fast computations and fast adaptivity in dynamic settings, and ensuring decision robustness under environment uncertainties.To develop such a comprehensive framework, this project has several research thrusts, each addressing one aspect of the framework. In each thrust, the project pursues the following two intertwined research objectives: (1) Theoretical Foundations: The first objective is to establish the fundamental limits of forming quickest situational awareness, while recognizing the physical restraints due to the cost of sensing, communication delay tolerance, model uncertainties, and dynamic variations of the beneficiaries of spectrum sensing, which encompass various autonomous wireless applications that operate under distinct protocols and infrastructures. (2) Algorithm Design: Driven by the observations gained from the uncovered fundamental limits, the second objective is to design data analytic tools and mechanisms for forming the quickest situational awareness, which is amenable to real-time and scalable implementation, exhibiting sound decision guarantees, and fully adaptive and responsive to the spatio-temporal variations of the spectrum occupancy status in wideband spectrum. This project will produce a decision-theoretic framework for forming quick and reliable situational awareness of the spectral occupancy states over different segments in wideband spectrum. To achieve this goal, the project will address long-standing open problems in quickest detection of change over networked data, open problems in controlled sensing, as well as challenging problems motivated by the physical constraints of spectrum sensing such as 1) sensing budget, 2) decision-making under model uncertainty, and 3) detecting adversaries that seek to compromise sensing data and distort spectrum sensing decisions.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.