CAREER: Automating the measurement and management of the radio spectrum for future spectrum-sharing applications

越来越多推动经济增长和人类福祉的领域，包括医疗保健、应急管理和国防，都依赖于无线网络连接。这创造了6409亿美元的市场潜力，这是现有网络满负荷运行时无法实现的。尽管有这种潜力，但只有8%的无线电频谱分配给了无线通信技术。这种最小的分配造成了频率资源的人为稀缺，从而使流行频段饱和，而其他频段则未得到充分利用。作为回应，无线技术已经开始整合新的硬件和软件，以通过机会主义的频率重用来提高其频谱效率。尽管前景看好，但这一创新轨迹无法持续，除非我们建立一个原则性频谱测量和管理框架，该框架可以包含不可预见的网络和传感器能力，以支持未来的频谱政策、监管和技术。该项目开发了一项长期、综合的研究、教育和推广计划，以(I)建立支持共享频谱接入的自动化频谱测量的科学和技术框架，以及(Ii)培训网络、数字通信和机器学习交叉领域的下一代无线专家。该项目将与工业和标准化工作密切合作，以确保更广泛的采用。研究将在信号处理、数字通信、机器学习、图形挖掘和大规模测量的交汇处进行。首先，它将研究真实世界扫描缺陷对信号特征的影响。根据这些见解，该项目将为频谱识别和来自不完美扫描的发射器指纹识别贡献算法。其次，它将创建一个分析框架，将传感器属性与数据质量和算法性能联系起来，从而实现应用程序驱动的测量。第三，模型和算法将集成在社区可访问的开放系统中。该项目将为面向未来和应用驱动的频谱测量奠定基础，该测量利用宽带和异质传感来支持新兴无线网络的端到端。它将有助于数据驱动和平台感知的抽象、模型和算法，为自动频谱分析产生领域信息特征。结果将弥合测量能力和算法要求之间的现有差距，同时处理流氓或不完全扫描的发射器的现实场景。该项目将产生一个成本效用框架，以探索传感器价格、带宽、计算和功率；以及算法的准确性、稳定性和置信度之间的权衡。将研究成果整合到一个开放的系统中，将允许系统和可重复的研究和教育，同时遵守并告知标准化工作。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

A growing number of domains that drive economic growth and humanity's well-being, including healthcare, emergency management and national defense, hinge on wireless network connectivity. This has created a market potential of $640.9 billion, which cannot be realized as existing networks operate at capacity. Despite this potential, only 8% of the radio spectrum is allocated to wireless communication technologies. This minimal allocation creates artificial scarcity of frequency resources, whereby popular bands are saturated, while others are under-utilized. In response, wireless technologies have begun to incorporate new hardware and software to boost their spectrum efficiency through opportunistic frequency reuse. While promising, this trajectory of innovation cannot be sustained unless we establish a framework for principled spectrum measurement and management that can embrace unforeseen network and sensor capabilities in support of future spectrum policy, policing and technology. This project develops a long-term, integrated program of research, education and outreach to (i) establish a scientific and technological framework for automated spectrum measurement in support of shared-spectrum access, and (ii) to train the next generation of wireless specialists at the intersection of networks, digital communications and machine learning. The project will work closely with industry and standardization efforts to ensure broader adoption.The research will be carried out in three thrusts at the confluence of signal processing, digital communications, machine learning, graph mining, and large-scale measurement. First, it will study the effects of real-world scan imperfections on signal features. Following these insights, the project will contribute algorithms for spectrum cognizance and transmitter fingerprinting from imperfect scans. Second, it will enable application-driven measurement by creating an analytical framework that links sensor properties with data quality and algorithm performance. Third, models and algorithms will be integrated in an open system accessible to the community. This project will lay the foundation for future-proof and application-driven spectrum measurement that leverages wide-band and heterogeneous sensing for end-to-end support of emerging wireless networks. It will contribute data-driven and platform-aware abstractions, models and algorithms that produce domain-informed features for automatic spectrum analytics. The outcomes will bridge the existing gap between measurement capabilities and algorithm requirements while tackling realistic scenarios with rogue or incompletely scanned transmitters. The project will produce a cost-utility framework to explore tradeoffs between sensor price, bandwidth, computation and power; and algorithm accuracy, stability and confidence. The integration of research outcomes into an open system will allow systematic and reproducible research and education, while complying with and informing standardization efforts.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.