Networking a dependable cyberinfrastructure

普通人的日常生活，从银行到购物，从学习到娱乐，越来越依赖于以互联网为中心的网络基础设施。在过去的十年中，大量的研究和投资致力于提高互联网的性能和可靠性，主要是在其主干上。然而，在接入网和互联网服务提供商之间提供这种可靠性仍然存在许多障碍。拒绝服务攻击、电子邮件垃圾邮件和网络诈骗让许多互联网用户（包括消费者和企业）措手不及。域名系统（DNS）、边界网关协议（BGP）和会话发起协议（Session Initiation Protocol）的控制平面漏洞更是令人担忧。现有的DNS Security、Secure BGP等方案复杂度较高，很少被部署。对于互联网这样一个复杂的系统，却没有嵌入系统的、在线的测试能力，这是非常令人惊讶的。总而言之，互联网的可靠性仍然远远低于它应有的水平，大多数人认为这是理所当然的。我们对可靠的网络基础设施的研究计划侧重于基本的网络问题，以大大提高互联网及其附件的端到端可靠性，安全性和可测试性。该研究探讨了三个高度相关的主题：(1)我们希望利用无线个人、本地、城域网甚至广域网在ad hoc或混合模式下的核心互联网的可靠性。(2)我们希望增强互联网控制平面的安全性，特别是在其命名、路由和信令基础设施方面。(3)我们希望将系统和在线可测试性嵌入到网络基础设施中，而不是作为另一个附加组件。这项研究也有望为学术界和工业界培养出高素质的人才，以应对互联网下一次重大转型期间的巨大挑战。

The daily life of ordinary people, ranging from banking to shopping and from learning to entertaining, is increasingly dependent on the Internet-centric cyberinfrastructure. In the last decade, a vast amount of research and investment has been devoted to improving the performance and reliability of the Internet, primarily in its backbone. However, there are still many obstacles left in providing such reliability in access networks and among Internet Service Providers. Denial-of-Service attacks, email spams and web scams have caught many Internet users, both consumers and businesses, off-guard. The control-plane vulnerabilities in Domain Name System (DNS), Border Gateway Protocol (BGP) and Session Initiation Protocol are even more alarming. Existing schemes such as DNS Security and Secure BGP have high complexity and are rarely deployed. For such a complex system as the Internet, it is very surprising to find that there are no systematic and online testing capabilities embedded. In summary, the Internet is still much less dependable than what it should be and most people have taken for granted. Our research program on a dependable cyberinfrastructure focuses on the fundamental networking issues in order to considerably improve the end-to-end reliability, security and testability of the Internet and its accessories. The research explores three highly correlated topics: (1) We want to leverage the reliability at the core Internet with that of wireless personal, local, metropolitan and even wide area networks in ad hoc or hybrid mode. (2) We want to enhance the control plane security for the Internet, particularly in its naming, routing and signaling infrastructures. (3) We want to embed the systematic and online testability into the cyberinfrastructure, rather than being another add-on. The research is also expected to produce highly qualified personnel for academia and industry with growing demands for such talent to meet this grand challenge during the next major transformation of the Internet.