CAREER: Unstructured Dynamic Overlay Networks and Strategic Users

该奖项是根据2009年美国复苏和再投资法案（公法111-5）资助的。该项目研究非结构化动态覆盖网络（非结构化网络）：一种由不断变化的战略（即，自私但理性的）用户。这些用户相互依赖来传输和查找数据，并且他们通过动态地引入/移除网络通信链路并在互联网上执行分布式资源分配来基于他们自己的成本效益权衡形成非结构化网络。 由于非结构化网络支持的应用程序非常普遍，因此非结构化网络承载了大部分Internet流量。 这样的应用包括文件共享（例如，BitTorrent）、用户辅助媒体流、视频点播和IP语音（例如，Skype）。这些网络还包括自私的数据路由覆盖（用户独立选择路由）和多个自组织的数据路由覆盖。每个非结构化网络内和网络之间的用户之间的基本交互/冲突，以及他们与互联网的交互，可能会导致非常不稳定的状态，低效的资源使用，他们支持的应用程序的低性能，以及对互联网服务提供商的不利影响。 为了解决这些问题，该项目正在开发有效的机制，以确保非结构化网络的稳定性和效率，并提供与互联网架构和服务提供商的无缝互操作。依靠优化和博弈论，随机建模和分布式算法/机制设计的组合，本研究开发的理论基础/方法，量化的互动动态内/之间的非结构化网络，非结构化网络和互联网之间。此外，该项目还为应用/网络设计人员和服务提供商提供了设计原则/算法，以满足日益丰富的用户需求，并将为未来互联网架构的设计提供指导。

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).This project investigates unstructured dynamic overlay networks (unstructured networks): a type of computer network formed from interactions among constantly changing strategic (i.e., self- interested but rational) users. These users rely on each other for transferring and finding data, and they form unstructured networks based on their own cost-benefit tradeoffs by dynamically introducing/removing network communications links and performing distributed resource allocations on top of the Internet. Unstructured networks carry the majority of Internet traffic due to the prevalence of the applications they support. Such applications include file sharing (e.g., BitTorrent), user-assisted media streaming, video-on-demand, and voice-over-IP (e.g., Skype). These networks also include selfish data-routing overlays (where users choose routes independently), and multiple self-organized data-routing overlays. The fundamental interactions/conflicts between users within each unstructured network and among networks, and their interactions with the Internet, can result in drastically unstable states, inefficient resource usage, low performance of the applications they support, and adverse impact on Internet Service Providers. To address these problems, the project is developing effective mechanisms to ensure the stability and efficiency of unstructured networks, and providing seamless interoperation with the Internet architecture and service providers. Relying on a combination of optimization and game theory, stochastic modeling, and distributed algorithm/mechanism design, this research develops theoretical foundations/methodologies for quantifying the interaction dynamics within/among unstructured networks, and between unstructured networks and the Internet. Additionally, the project provides design principles/algorithms for application/network designers and service providers for meeting increasingly rich user requirements and will provide guidance for the design of future Internet architecture.