TF:SING: Exploring the Fundamental Limits of Cross-Layer Optimization in the Next-Generation Internet

虽然已经有相当多的研究使用了跨层优化，但对跨层优化的设计极限仍然缺乏基本的认识。这个项目的首要目标是填补这一知识空白。目标是：1)定性表征跨层优化的多尺度时空行为；2)开发一种新的复杂性可扩展框架，以支持具有不同计算能力的异构网元的跨层优化；3)通过广泛的仿真和基于现有试验台的现场测试来评估所提出的框架。该项目的智力价值和变革意义包括：1)提供了一个理论框架来理解跨层设计和优化中的主要设计权衡；2)提供了一个共同的基础来定量评估和比较不同的跨层设计方案。通过开发一个新的理论框架来定量分析复杂系统优化中的设计权衡，该项目的更广泛影响是：1)有利于社会网络，经济学，物理学和生物学的广泛应用，从而对所有社会层面产生深远影响；2)扩大与工业伙伴的现有合作，并加强STEM教育的持续努力。结果、成果、软件工具、基准和教育材料将通过项目传播？网站以及期刊和会议出版物。跨层优化的研究成果也将被整合到PI所在大学的课程中。

Although there has been a considerable amount of research done by using cross-layer optimization, fundamental understandings on the design limits of cross-layer optimization are still missing. The overarching goal of this project is to fill this knowledge gap. The objectives are to 1) qualitatively characterize multi-scale temporal-spatial behaviors of cross-layer optimization; 2) develop a novel complexity-scalable framework to support cross-layer optimization over heterogeneous network elements with different computation capabilities; and 3) evaluate the proposed framework by using extensive simulations as well as field test based on our existing testbed.The intellectual merits and transformative significance of this project include: 1) to provide a theoretical framework to understand major design tradeoffs in cross-layer design and optimization and 2) to provide a common ground to quantitatively evaluate and compare different cross-layer design schemes. By developing a new theoretical framework to quantitatively analyze design tradeoffs in complex system optimization, the broader impacts of this project are: 1) to benefit wide applications in social networks, economics, physics, and biology, leading to a profound impact on all societal levels and 2) to expand the existing collaborations with industrial partners and enhance the ongoing effort on STEM education. Results, outcomes, software tools, benchmarks, and educational materials will be disseminated through a project?s web site as well as journal and conference publications. The research results of cross-layer optimizations will also be integrated into courses taught in the PI's university.