Back to search

Georgia Institute of Technology

Understanding and Improving Internet Censorship Measurement

Abstract

dc:description.abstract

Internet measurement methods over the last decade have enabled researchers to study critical problems such as understanding botnets, performing censorship measurement, and analyzing vulnerabilities. While these studies have primarily focused on improving scale and speed, crucial elements of reliability and correctness are comparatively neglected. Several studies have observed non-determinism or inconsistency in their results, but these effects remain largely unexplored. In this thesis, we introduce empirical methods and systems to evaluate factors that impact the reliability of Internet censorship measurements, showing that routing paths, packet parameters, measurement direction, and IP version all shape what is observed. Internet censorship is widespread, impacting citizens across many countries worldwide. Significant prior work has developed techniques to remotely perform large-scale, longitudinal measurements of global Internet manipulation. However, this work has largely prioritized scale over reproducibility and consistency. First, we study the causes of variation in external DNS censorship measurement in China. We show that ECMP routing, influenced by changes in packet header fields (e.g., source IP and source port), has a direct impact on censorship results. We find that 37% of IPs across 56% of ASes exhibit changes in observed censorship behavior depending on packet parameters, and that the magnitude of censorship differences can more than double based on the lowest three bits of the source IP, consistent with known router load balancing techniques. Second, we extend this analysis globally across protocols and censorship mechanisms. We show that routing-induced variation significantly affects remote censorship measurement in 17 of 21 countries studied, across DNS, HTTP, and HTTPS. We find that these differences arise from multiple sources, including intra-AS and inter-AS routing differences, path diversity to the same destination, and geographically non-uniform censorship deployment, demonstrating that the problem is not unique to China or a single protocol. Finally, to understand whether these effects depend on measurement direction or IP version, we develop a unified measurement framework that incorporates inside-out and outside-in measurements across both IPv4 and IPv6. We conduct a global study across 24 countries and show that remote measurements can observe disruption in 21 countries without requiring in-country resources. We find that IPv6 disruption is less prevalent than IPv4, though still present in countries such as China, Russia, and Iran. Critically, routing- and parameter-induced variation persists across both vantage models and IP versions, and externally observed blocking can arise from on-path behavior not visible at the destination. Together, these results demonstrate that Internet censorship measurement outcomes are strongly shaped by methodological choices. Measurement direction, IP version, and route-sensitive source parameters can all influence observed results, leading to variation in censorship observations across measurements of the same destinations. These findings highlight the need to explicitly control and account for these factors in order to obtain reliable and consistent measurements.

Degree

thesis:*
Name thesis:degree_name
Computer Science, PhD
Department thesis:degree_department
Computer Science
Grantor
Georgia Institute of Technology
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Bhaskar, Abhishek Vasisht
Advisor dc:contributor.advisor
  • Li, Frank
Committee members dc:contributor.committeemember
  • Dainotti, Alberto
  • Bailey, Michael
  • Specter, Michael
  • Pearce, Paul
ORCID person:identifier.orcid

Rights

Language dc:language.iso
English

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/1853/81734

Chain of custody

source
Harvested from
Georgia Tech
Base URL
repository.gatech.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Bhaskar, Abhishek Vasisht. Understanding and Improving Internet Censorship Measurement. Georgia Institute of Technology, 2026. https://hdl.handle.net/1853/81734