EARS: Collaborative: Comprehensive Network State Inference for Robust and Policy-Cognizant Spectrum Access

目标：这个项目的目标是一个简明的认知基础设施，能够在治理策略约束下进行健壮的、跨层的协议设计，从而最大化认知无线电(CR)网络的性能和终端用户的满意度，尽管存在与动态和机会主义的无线电频谱接入相关的挑战，以及反映获取网络状态测量的成本和限制的不完整、损坏和零星的数据。智力的优点：智力的优点是改变最先进的学习工具，用于全面地和资源感知地认知“全局”网络状态(包括干扰、任意到任意链路的增益、带宽占用、队列长度、和路径延迟)，以及具有政策含义的随机、状态认知的CR网络优化，开创了无线CR网络的路由、介质接入和物理层设计的变革性适应方法，赋予了对状态不确定性亟需的稳健性。广泛的影响：更广泛的影响包括对移动自组织、智能电网、智能交通网络以及医疗遥测、地理监测和监视系统的有形影响。在基于核的学习、核矩阵补全、图上的推理和基于采样的场景优化的基础上取得的进展将使包括社会分析、生物信息学、医学成像和使用传感器网络的监控在内的一系列研究领域受益。与技术部分相结合的政策研究将鼓励社区接受新的认知无线电技术，并打开投资和商业机会，产生更大的经济影响。由拟议的认知频谱操作测试平台推动的本科生设计项目和Botball竞赛也将有利于学生在最先进的学习、移动机器人上的无线系统和网络优化科目方面的实践经验培训。

Objective: The objective of this project is a compendious cognition infrastructure enabling robust, cross-layer protocol designs under governance policy constraints, so as to maximize the cognitive radio (CR) network performance and end-user satisfaction, notwithstanding the challenges associated with dynamic and opportunistic radio-spectrum access, as well as the incomplete, corrupt and sporadic data reflecting the cost and restrictions in acquiring network state measurements.Intellectual merit: The intellectual merit is to transform state-of-the-art learning tools for comprehensive and resource-aware cognition of the "global" network state (which includes interference, any-to-any link gains, band occupancies, queue lengths, and path delays), jointly with stochastic, state-cognizant optimization of CR networks with policy implications, to pioneer transformative adaptation approaches to routing, medium-access, and physical-layer designs of wireless CR networks endowed with the much needed robustness to state uncertainty.Broader impacts: The broader impacts include tangible implications to mobile ad hoc, smart grid, intelligent transportation networks, as well as medical telemetry, geo-monitoring, and surveillance systems. Advances in the foundations of kernel-based learning, kernel matrix completion, inference on graphs, and sampling-based scenario optimization, will benefit a gamut of research areas including social analytics, bio-informatics, medical imaging, and surveillance using sensor networks. Policy research integrated with technical components will encourage community embracing of novel CR technologies and open up investment and business opportunities yielding greater economic impact.Undergraduate Design Projects and Botball Competition, facilitated by the proposed Cognitive Spectrum Operations Testbed, will also benefit student training with hands-on experience in state-of-the-art learning, wireless systems on mobile robots, and network optimization subjects.