CIF:Small: Covariance Arbitrage: Unlocking the Full Value of Correlation Diversity in Multiuser Wireless Networks

在许多新兴的无线通信技术中，无线信道状态信息的获取和传输已经发展成为系统设计和运行中的首要问题。一个典型的例子是大规模MIMO，特别是分频双工。这需要仔细处理用于训练和数据传输的通道资源。该项目致力于仔细分析涉及信道空间相关性的现象，并从中获益的新解决方案的综合，从而实现更有效的信道训练和传输，特别是在上述效率可能对无线系统性能产生关键影响的情况下。这项研究还辅之以教育和外联活动，包括培训研究生和本科生。在大规模MIMO中，物理信道的散射特性会产生秩不足的协方差矩阵，这为重新获得信道训练效率的一些有趣建议打开了大门，特别是使用中长期协方差特征空间进行预波束形成。除了大规模MIMO之外，无线节点协方差特性之间的变化也会发生在其他场景中。所提议的活动致力于在无线网络中建立相关多样性理论，其思想很像交易者的套利，可以从协方差特征空间的差异中获利，无论是通过重叠还是通过特征空间的分离。所提议的活动旨在开辟比目前基于协方差的技术更广泛的适用性领域，这些技术的最佳增益是在特定的天线配置、接收器数量和协方差等级行为下实现的。所提出的活动也与在时域和频域中发现的密切对应联系在一起，为协方差和相干差异的综合理论提供了一个突破口。

In many emerging wireless communication technologies, the acquiring and transport of information about the state of the wireless channels has evolved into a first-order issue in the design and operation of the system. A prime example is observed in massive MIMO, especially the frequency-division duplex variety. This requires careful handling of channel resources that are used for training versus data transmission. The project is dedicated to a careful analysis of phenomena involving channel spatial correlations and synthesis of novel solutions that benefit from them, leading to more efficient channel training and transmission especially in scenarios where said efficiency can have a critical impact on wireless system performance. The research is complemented by educational and outreach activities, including training of graduate and undergraduate students.In massive MIMO, scattering properties of physical channels produce rank-deficient covariance matrices, opening the door to a number of interesting proposals for recapturing efficiency of channel training notably to use medium- to long-term covariance eigenspaces for pre-beamforming. The variation between covariance properties of wireless nodes also occurs in scenarios other than massive MIMO. The proposed activity is dedicated to the building of a theory of correlation diversity in wireless networks animated by the idea that, much like a trader's arbitrage, it is possible to profit from differences of covariance eigenspaces, either via overlap or via separation of eigenspaces. The proposed activity aims to open a wider domain of applicability than present covariance-based techniques whose best gains are under certain antenna configurations, number of receivers, and covariance rank behavior. The proposed activity also ties in with close counterparts that are found in the time- and frequency-domain, providing an opening to a comprehensive theory of covariance and coherence disparity.