CAREER: Theoretical Foundations for Wireless Network Algorithm Design: Satisfying Short-Term and Long-Term Application Requirements

设想构成许多未来通信系统骨干的多跳无线网络通常受到限制性和不可预测变化的资源约束；要求使用低复杂性和可扩展的操作进行操作；设想为具有一系列长期（例如，吞吐量贪婪，长时间应用程序）到短期（例如，延迟敏感，短时间应用程序）需求的基本应用程序提供服务。然而，对于如何设计在如此恶劣的条件下运行的无线网络控制器，以有效地服务于各种应用，特别是那些具有不同程度延迟敏感性的应用，我们的理解存在巨大差距。本研究通过为有效支持通用多跳无线网络中各种应用的通信策略的系统设计提供严格的理论基础，填补了这一空白。为此，本文首先揭示了几个未探索的设计选择对实现的应用程序性能的重要影响，包括服务规程、流量塑造策略和延迟感知决策制定。然后，通过在非凸优化、随机网络控制和随机/近似算法中开发新的方法和数学工具，系统地利用这些新的设计维度，以最佳地满足与延迟相关的短期和与吞吐量相关的长期应用需求。在这个项目中开发的见解和数学工具预计将在设计未来网络的有效通信协议方面发挥重要作用，这些网络服务于具有一系列延迟敏感性和吞吐量要求的重要服务（在医疗保健、救援行动、安全、自动控制等方面）。社会也将受益于该项目的教育方面的培训未来的工程师和研究人员。

Multi-hop wireless networks that are envisioned to form the backbone of many future communication systems are: typically subject to restrictive and unpredictably changing resource constraints; required to operate using low complexity and scalable operations; envisioned to serve essential applications with a range of long-term (e.g. throughput-greedy, long-duration applications) to short-term (e.g. delay-sensitive, short-duration applications) requirements. Yet, there is a huge gap in our understanding of how to design wireless network controllers operating under such harsh conditions that can efficiently serve diverse applications, especially those that have various degrees of delay sensitivities.This research fills this gap by developing a rigorous theoretical foundation for the methodical design of communication strategies for effectively supporting diverse applications in general multi-hop wireless networks. To that end, it first reveals the significant impact of several unexplored design choices, including the service discipline, traffic shaping strategy, and delay-aware decision making, on the achieved application performances. Then, it systematically exploits these new design dimensions for optimally serving both delay-related short-term and throughput-related long-term application requirements by developing new approaches and mathematical tools within non-convex optimization, stochastic network control, and randomized/approximate algorithms. The insights and mathematical tools developed within this project are expected to play an instrumental role in the design of efficient communication protocols for tomorrow's networks serving vital services with a range of delay sensitivities and throughput requirements (in health-care, rescue operations, security, automated control, etc.). The society will also benefit from the educational aspects of the project in the training of future engineers and researchers.