82. CHARACTERIZING RESIDENTIAL BROADBAND PERFORMANCE IN THE UNITED STATES: INFERRING CONGESTION ON INTERDOMAIN LINKS
Name: Alexander Manuel Gamero Garrido
Grad Year: 2020
We aim to illuminate tensions between Internet service providers (ISPs) by providing data on the existence of persistently congested inter-domain links in the United States; and by creating a new congestion map of the Internet, with particular emphasis on interconnection links involving residential providers. Combined, these findings will: -Provide evidence regarding the health of residential connectivity. -Identify any actors (or groups of actors) persistently running congested links. -Identify and characterize events of high congestion across providers. -Explore the impact of interdomain congestion on end-to-end latency, a leading determinant of end-user quality of service. This project's broad scope appeals to the interests of at least three groups. First, individual consumers and their advocacy organizations, concerned with events impacting performance of their broadband connection. Second, federal regulators, interested in preventing anti-competitive practices and consumer harms. Third, the Internet measurements research community, which would gain access to our datasets collected from a distributed, nation-wide network. Finally, though broadband and content providers have access to private, massive datasets already, several of them have agreed to provide us with data to confirm our inferences. Our exploration of inter-domain congestion faces two main challenges. First, the nature of tussles between Internet service providers (ISPs) is not well understood, given that non-disclosure agreements (NDA) protect peering contracts. And second, there is insufficient network measurements data available to study interconnection disputes. We use a combination of topology probes and latency measurements from residential vantage points to circumvent these challenges. Specifically, we deploy measurement software to access network nodes, in order to both detect inter-domain congestion and characterize its potential impact on Quality of Service. Additionally, we leverage data analysis techniques on existing latency data collected using the CAIDA Ark infrastructure, combined with topological data, to develop a congestion "heat map" of the Internet.
Industry Application Area(s)
Internet, Networking, Systems | Software, Analytics | Public Policy Consulting