CAREER: Scalable Distributed MIMO: Towards Density-Proportional Capacity Scaling for Infrastructure Wireless Networks

移动设备的激增推动了对基础设施无线网络容量的不断增长的需求，尤其是Wi-Fi。目前的无线网络很难满足这一需求，因为它们是干扰有限的--每个用户的吞吐量随着网络密度超线性下降。添加更多接入点(AP)可能会使问题恶化，因为AP本身会相互干扰。本研究的目的是通过探索一种称为可扩展分布式MIMO(SDM)的变革性架构，来克服这一基本限制，并使网络容量能够与节点密度成比例地扩展。SDM背后的原则是将AP重组为集群。在每个群集中，AP彼此紧密同步和共享数据。这使他们能够消除相互干扰，并根据AP密度扩展网络容量。不同的AP群集以自组织的方式竞争通道访问，从而扩展整个网络的容量。SDM的长期目标是实现密集的室内基础设施网络，支持每用户Gbps的吞吐量，标志着朝着国家宽带计划中概述的目标迈出了重要的一步。为了实现SDM原则，本研究综合了一个跨越五年的综合任务框架，包括性能建模/分析、网络协议/算法设计以及在软件无线电试验台上的实现/实验。特别是，PI计划(I)系统地探索部署AP集群的方法，以平衡系统复杂性、可扩展性和与传统网络的兼容性之间的权衡；(Ii)设计以集群为中心的网络协议的新范例，以集群内AP之间的紧密协调和不同集群之间的自组织为特征；(Iii)开发新的通信算法，在最大限度地提高集群容量的同时驯服协调开销，并将容量收益转化为改善的最终用户体验。研究任务中使用的试验床将扩展为一个用户友好的教育平台，为不同级别的学生提高无线网络知识。研究和教育材料将广泛传播，以便于结果的重现性。

The proliferation of mobile devices is driving an ever-increasing demand on the capacity of infrastructure wireless networks, especially Wi-Fi. Current wireless networks can hardly meet this demand because they are interference limited -- per-user throughput decreases super-linearly with network density. Adding more access points (APs) may worsen the problem, because the APs themselves interfere with each other. The objective of this research is to overcome this fundamental limitation, and enable network capacity to scale proportionally with node density, by exploring a transformative architecture called Scalable Distributed MIMO (SDM). The principle behind SDM is to reorganize the APs into clusters. Within each cluster, the APs tightly synchronize and share data with each other. This enables them to cancel the mutual interference and scale network capacity with AP density. Different AP clusters contend for channel access in a self-organized manner, thus scaling capacity across an entire network. The long-term objective of SDM is to enable dense indoor infrastructure networks supporting Gbps per-user throughput, marking an important step towards the goals outlined in the National Broadband Plan. To realize the SDM principle, the proposed research synthesizes a comprehensive framework of tasks spanning five years, including performance modeling/analysis, network protocol/algorithm design, and implementation/experimentation on a software radio testbed. In particular, the PI plans to (i) systematically explore ways of deploying AP clusters to balance the tradeoffs between system complexity, scalability, and compatibility with legacy networks; (ii) design a new paradigm of cluster-centric network protocols that feature tight coordination between APs within a cluster and self-organization among different clusters; (iii) develop new communications algorithms that tame the coordination overhead while maximizing cluster capacity, and translate the capacity gain into improved end-user experience. The testbed used in the research tasks will be extended into a user-friendly educational platform that enhances the knowledge of wireless networks for students at different levels. The research and educational materials will be broadly disseminated for reproducibility of results.