Internet-Wide Measurements

Understanding Internet behavior requires empirical measurement at scale rather than relying solely on models and simulations. Real networks exhibit complexity that theoretical analysis struggles to capture: middleboxes modify traffic in deployment-specific ways, routing policies reflect business relationships rather than shortest paths, performance varies with geographic location and time of day, and new protocols and technologies deploy unevenly across the global Internet. Measurement reveals the gap between how systems are designed to work and how they actually behave in practice—a gap that often contains the most important insights for protocol design, deployment strategies, and system optimizations. Large-scale measurement campaigns spanning diverse geographic regions, network types, and vantage points are essential to characterize protocol adoption rates, infrastructure performance characteristics, service availability patterns, and to validate whether theoretical benefits of new technologies materialize in real deployments.

Our research conducts large-scale measurements to characterize Internet infrastructure and protocol behavior across diverse conditions. We deploy measurement campaigns spanning multiple continents and network types to understand how LEO satellite networks perform compared to terrestrial infrastructure, revealing performance characteristics, CDN caching behaviors, and protocol adaptation requirements. We analyze protocol adoption patterns and deployment barriers for new transport protocols, measuring where theoretical benefits materialize and where real-world constraints—middleboxes, software update cycles, interoperability issues—prevent adoption. Our measurements characterize CDN performance across different network types, infrastructure reachability and availability patterns, and routing behavior under various conditions. We leverage global measurement platforms including RIPE Atlas (12,000+ worldwide probes) and deploy custom instrumentation for specialized measurements requiring detailed data collection beyond standard probe capabilities. Measurement methodologies combine active probing (controlled traffic injection to test specific behaviors) and passive monitoring (observing real traffic patterns) depending on research questions. Our measurement infrastructure supports privacy-preserving data collection complying with regulations while enabling valuable research. We release open datasets and measurement tools to enable reproducible research by the broader community. Our work informs protocol standardization at IETF, guides network operators' deployment decisions, and has been featured in media highlighting Internet infrastructure's impact in underserved regions. Research directions include sustainable long-term measurement infrastructure, privacy-preserving measurement techniques, representative sampling across diverse populations, and real-time analysis capabilities for operational monitoring.