CAREER: Robust, Stable and Secure Routing via a vertically integrated monitoring and introspection system

PI的长期职业目标是帮助将互联网转变为强大、稳定和安全的路由基础设施，以提供高度可靠和可预测的性能。考虑到互联网在我们日常生活中扮演的关键角色，当前的路由体系结构令人惊讶地脆弱。光纤切断、设备故障或配置错误以及恶意攻击(例如Nimda蠕虫)已导致全球连接的广泛丢失。为了实现我们的愿景，我们制定了以下四个研究议程：1.PI认为，一个重要的起点是彻底描述广域故障场景以及它们对流量转发平面的影响程度。通过PI与Sprint的合作，将监控运营的Tier-1互联网服务提供商的主干，以收集路由和流量数据。基于对数据的连续调节，将推导出广域故障(WARF)模型，该模型可被研究团体用来在模拟或试验台环境中产生现实故障。PI将进行补充工作，设计一个统计BGP异常检测器，该检测器自动执行区分异常和预期路由行为的过程。该设计在创建历史配置文件和执行短期测试时遵循基本的入侵检测原则。3.PI将通过设计和开发促进快速路由收敛和流量工程的覆盖策略控制体系结构(OPCA)来研究策略路由的替代方法。OPCA将允许在域内和域间路由中同时使用多种类型的度量，并说明这种功能在IP核中是必要的。PI将设计一个路由自省和反馈系统(RIF)，向覆盖网络和传输层代理或应用层代理等高层实体提供及时的反馈。将扩展我们的研究，探索一种混合信道编码和重传方案，以基于故障检测和路由环路来优化视频流。这项工作的成果是使核心互联网更加可靠和稳定，从而使基于互联网的分布式计算成为可能。跨异类网络(包括无线和卫星网络)的有效路由是实现真正无处不在的连接的关键，这将对社会应用产生广泛影响。本项目所验证的假设和方法将有助于增进广域布线领域的知识，并为分析其他大规模分布式系统的可靠性问题奠定基础。将使在此项目中开发的故障模型、工具和原型可供其他研究人员在线使用，并通过出版物和培训研讨会提供。PI的教育使命是通过将现实生活中的运营经验和广域互联网行为的轶事融入课堂教学，培养本科生和研究生成为有能力的网络工程师。这包括开发一门顶石设计课程，为学生提供网络管理和路由模块的实践体验。PI将通过与加州大学戴维斯分校的女性工程学(WIE)和数学、工程和科学成就(MESA)项目合作，积极招募代表性较低的女性和少数族裔学生进入工程课程及其研究项目。

The long-term career goal of the PI is to help transform the Internet into a robust, stable, and secure routing infrastructure that delivers highly reliable and predictable performance. Considering the critical role that the Internet plays in our day-to-day lives, the current routing architecture is surprisingly fragile. Fiber cuts, faulty or mis-configured equipment, and malicious attacks (e.g., the Nimda worm) have led to a widespread loss of global connectivity. To achieve our vision, we have developed the following four research agendas: 1. PI argue that an essential starting point is a thorough characterization of wide-area failure scenarios and how much they impact the traffic-forwarding plane. Through the PI's collaboration with Sprint, will monitor an operational Tier-1 Internet Service Provider's backbone to collect routing and traffic data. Based on successive conditioning of the data, will derive a wide-area failure (WARF) model that can be used by the research community to generate realistic failures in simulation or testbed environments.2. PI will undertake a complementary effort to design a statistical BGP anomaly detector that automates the process of differentiating abnormal and expected routing behavior. The design follows basic intrusion detection principles in creating a historical profile and performing short-term testing. 3. PI will investigate an alternative approach to policy routing by designing and developing an Overlay Policy Control Architecture (OPCA) that facilitates fast route convergence and traffic engineering. OPCA will allow the concurrent use of multiple types of metrics in intra- and inter-domain routing, and illustrate that such functionality is warranted in the IP core.4. PI will design a Routing Introspection and Feedback System (RIFS) that provides timely feedback to higher-layer entities such as overlay networks and transport or application layer proxies. will extend our study to explore a hybrid channel coding and retransmission scheme to optimize video streaming based on detection of failures and routing loops. The outcome of this work strives to enable Internet-based distributed computing by making the core Internet more reliable and stable. Effective routing across heterogeneous networks (including wireless and satellite) is the key to truly ubiquitous connectivity, which will have a broad impact on societal applications. The hypothesis and methodologies tested in this project will help advance the knowledge in the field of wide-area routing and form a foundation for analyzing the reliability issues of other large-scale distributed systems. Will make the failure models, tools, and prototypes developed in this project available to other researchers online and via publications and training workshops. The PI's educational mission is to train undergraduate and graduate students to become capable network engineers by incorporating real-life operational experience and anecdotes of wide-area Internet behavior into classroom teaching. This includes developing a capstone design course that gives students hands-on experience in network management and routing modules. The PI will actively recruit underrepresented women and minority students into the engineering curriculum and her research projects by collaborating with the Women in Engineering (WIE) and the Mathematics, Engineering, and Science Achievement (MESA) programs at U. C. Davis.