A central premise of my work is that a meaningful understanding of BGP requires understanding BGP communities. Communities encode much of the routing policy, operational intent, and location information that govern how routes are propagated and selected in practice, yet they remain largely opaque: the vast majority of communities observed in the wild lack public documentation.

I develop models that extract structure and meaning from undocumented BGP communities, enabling inference of routing intent and the interaction between logical connectivity and physical geography. By making these signals interpretable, my research aims to explain how policy and deployment decisions shape the robustness of Internet connectivity.

Beyond routing policy, I also investigate the deployment and effectiveness of key Internet security mechanisms, including RPKI and DDoS protection services, with the goal of understanding their real-world adoption, operational trade-offs, and impact on Internet resilience.

To support this work, I design and operate measurement infrastructure that broadens visibility into Internet routing and enables large-scale, longitudinal, and reproducible analysis of BGP behavior. This includes community-facing services such as BGP2GO, which lowers the barrier to routing analysis by allowing users to efficiently locate relevant BGP update data without downloading large volumes of raw measurements.

This research agenda develops along two main directions: interpreting routing policy signals encoded in BGP communities, and expanding the measurement infrastructure used to study Internet routing behavior.

Understanding Internet Routing through BGP Communities

Much of the operational logic of Internet routing is encoded in BGP communities, which network operators use to signal routing policy, operational intent, and geographic location. In the IMC paper “Coarse-grained Inference of BGP Community Intent” (Best Paper Runner-up), I demonstrated how coarse-grained routing policies encoded in communities can be systematically interpreted using large-scale BGP measurements and operator-documented community semantics, building a cornerstone for subsequent finer-grained inference methods.

Building on this foundation, my more recent work investigates how BGP location communities can be used to infer where networks interconnect geographically. This includes the CoNEXT paper “Towards Understanding City-Level Routing Using BGP Location Communities” and my ongoing APNIC- funded project “Mapping Internet Peering Locations for Resilience and Policy in the Asia-Pacific.” Together, this line of work aims to reveal the geographic and policy-shaped structure of Internet connectivity using signals embedded in public routing data.

Measurement Platforms for Internet Infrastructure Research

A second research direction focuses on improving the infrastructure used to collect BGP data for Internet measurement. In the SIGCOMM Best Paper “The Next Generation of BGP Data Collection Platforms,” I contributed to the design of a novel sampling strategy for selectively collecting BGP updates from an order of magnitude larger set of BGP peers than traditional systems. This significantly improves visibility into global routing dynamics and helps mitigate the long-standing hidden-link problem in BGP measurements.

Another effort in this direction is BGPFiend, a modular framework for exploring and analyzing BGP data and BGP communities. This work will be presented as an INFOCOM poster titled “Modernizing BGP Data Access with BGPFiend.” The goal of this line of work is to modernize the infrastructure used in Internet measurement and enable researchers to more easily build datasets and analyses of global routing behavior.