Route Diagnosis in Path Vector Protocols

In this paper we present a novel approach to route diagnosis for path-vector routing protocols such as the Internet’s Border Gateway Protocol (BGP). Given a sequence of routing updates, our objective is to understand why a previous path was removed, why the specific new path is chosen among all existing alternatives, and whether any of the routers along the path injected false information. We collect necessary topological information for route diagnosis by inferring topological connectivity from routing updates, enhancing the path vector protocol with root cause notification, and using active queries between neighbor routers to learn missing information when necessary. In the absence of false routing updates, our design can reason about route removals and replacements with 100% accuracy. We also use majority votes among independent information sources, combined with active queries, to identify the source of false routing updates. Simulation results show that, in the presence of false updates, our design can achieve high detection rate and attacker identification rate with a low overhead.