EARS: Machine Learning and Social Protocols for Enhancing Spectrum Access for Wireless Communications

无线电频谱是一种需要管理的稀缺资源。为此，本项目设计、实施和研究新的协议，利用人类行为的基本原理来增强对无线电频谱的访问。它为有望显著提高频谱接入效率的协议提供技术、理论和指南。这可能会带来巨大的社会影响，因为它允许用同样的资源完成更多的工作。通过限制所需数据流量所需的基础设施，成功还将有利于环境；能源和基础设施投资都可以最小化。此外，该项目在特定的技术环境中创建了人类行为的模型，该模型可以作为涉及资源优化的其他技术领域的类似项目的基础。该项目创建了匿名数据集，以及分析人类和无线行为的工具，所有这些都将分发给公众。该项目在高中、本科和研究生阶段具有重要的教育和培训效益，包括本科研究原型项目，以及新的研究生水平课程和研讨会。这个跨学科项目应用并开发了来自社会计算、机器学习、无线技术、安全工程、物理模拟、移动系统和以用户为中心的设计等领域的专业知识。该项目实现并部署了一个系统，该系统可以通过机器学习和社交协议实现高效的带宽共享，这超出了仅凭技术所能实现的范围。社交协议是一种合作而又自由裁量的方法，它允许用户使用固有的自然决策过程（而不是外部强加的决策过程）更公平地分配访问权限。该项目包括三个主要活动：(1)研究社会协议的主要方法，包括说服性计算、临床行为改变理论和微任务；(2)开发一个系统，观察试图访问网络的用户，并制定控制规则，以最透明的方式让用户获得最大价值；(3)在实时网络中部署这些社交协议和系统，以评估现实环境中的方法。

Radio spectrum is a scarce resource that needs to be managed. Towards this end, this project designs, implements and studies novel protocols for enhancing access to radio spectrum by taking advantage of the fundamentals of human behavior. It produces technologies, theories, and guidelines for protocols that are expected to significantly improve the efficiency of spectrum access. This may lead to substantial societal impact by allowing more work to be done with the same resources. Success will also benefit the environment by limiting the infrastructure needed for the required data traffic; both energy and infrastructure investment can be minimized. Further, the project creates a model of human behavior in a specific technical context that can serve as a basis for similar projects in other technology areas involving resource optimization. The project creates anonymized datasets, and tools to analyze human and wireless behavior, all of which will be distributed to the general public. This project has important educational and training benefits at high school, undergraduate and graduate levels, including undergraduate research prototyping projects, and new graduate-level courses and seminars. This interdisciplinary project applies and develops expertise from areas of social computing, machine learning, wireless technology, security engineering, physical analogs, mobile systems, and user-centered design. The project implements and deploys a system that enables efficient bandwidth sharing with machine learning and social protocols that goes beyond what is possible with technology alone. Social protocols are cooperative yet discretionary methods that allow to users distribute access more fairly using inherently natural decision-making processes as opposed to externally imposed ones. The project consists of three major activities: (1) study the main approaches to social protocols, including persuasive computing, clinical behavior change theories, and micro-tasks; (2) develop a system that observes users trying to access the network and facilitates control rules that allow maximum value to users in the most transparent way; and (3) deploy these social protocols and the system in live networks to evaluate the approaches in real-life settings.