CIF: Small: Fundamental Limits of Wireless Communications with Stochastic Information Flows and Queueing Constraints

移动数据流量最近经历了前所未有的增长，预计未来几年还会进一步增长。移动数据和多媒体内容流量的这种指数增长对无线网络产生了重大影响。首先，这种无线多媒体业务需要一定的服务质量保证。另一个后果是网络流量的异构性。无线网络现在在不同的环境中承载着不同的随机业务，而移动多媒体通信网络的成功设计和有效的服务质量提供关键取决于对源业务模型的适当选择。在这些因素的推动下，该项目解决了具有服务保证的无线通信这一重要而及时的主题。这项研究的结果将对下一代无线网络的设计有重要的贡献，支持多媒体流量的服务保证，并将由于移动视频流量的极快增长而对社会产生重大影响。本项目的方法论集中在结合信息论和随机网络演算的工具，通过严格确定排队约束下无线通信的基本极限来建立一个强大的分析框架。特别是，这个项目调查了在随机数据到达和统计排队约束存在的情况下，单用户和多用户无线信道的最大吞吐量。首先，讨论了一种理想的设置，该设置具有高斯码本和没有码长限制且没有错误的可靠通信。随后，无线吞吐量被描述为更实用的有限块长度编码和具有任意分布的有限字母表输入的区域。

Mobile data traffic has experienced unprecedented growth recently and is predicted to grow further over the coming years. This exponential growth in the flow of mobile data and multimedia content has significant implications on wireless networks. For one, such wireless multimedia traffic requires certain quality of service guarantees. Another consequence is heterogeneity in network traffic. Wireless networks now carry heterogeneous stochastic traffic in diverse environments, and successful design of networks and effective quality of service provisioning for mobile multimedia communications critically depend on the appropriate choice of source traffic models. Motivated by these factors, this project addresses the important and timely topic of wireless communication with service guarantees. The outcomes of this research will significantly contribute to the design of next generation wireless networks supporting multimedia traffic with service guarantees, and are poised to have significant impact on the society due to extremely fast growth in mobile video traffic.The methodology of this project is centered around combining tools from information theory and stochastic network calculus to establish a strong analytical framework by rigorously determining the fundamental limits of wireless communication under queuing constraints. In particular, this project investigates the maximum throughput in single- and multi-user wireless channels in the presence of random data arrivals and statistical queuing constraints. First, an idealistic setup with Gaussian codebooks and reliable communication with no bounds on the code lengths and no errors is addressed. Subsequently, wireless throughput is characterized in the more practical regimes of finite block-length coding and of finite-alphabet inputs with arbitrary distributions.