Interdomain routing characterization and modeling for traffic engineering and security

Abstract

The BGP protocol has built-in limitations that jeopardize the Internet's fundamental goal of data delivery. First, BGP's information-hiding approach significantly complicates collaboration between autonomous systems, even in cases where there is mutual interest. Second, the absence of authentication and security primitives allows attacks against the Internet's interdomain routing system and causes simple configuration errors to disconnect millions of users. In this project, we will approach these built-in limitations in a three-step approach. First, we will develop new measurement and monitoring techniques to characterize interdomain routing on the Internet with greater precision, deepening our understanding of Internet properties. In particular, we will build a public database of BGP communities and will explore alternative routes to build a more complete map of the Internet's topology. Second, we will apply this improved understanding in building a new, more representative model of interdomain routing on the Internet. We will develop a model to predict Internet routes with greater accuracy than existing approaches and to estimate announcement configurations that lead to routes with given properties. Third, we will apply our model and its predictions in traffic engineering and anomaly detection solutions. The goals of these projects will benefit the research community, making available new databases and models, the network operator community, easing traffic engineering and anomaly detection, and Internet users, indirectly improving Internet performance and dependability. (AU)