ProtoScan: Measuring censorship in IPv6

Internet censorship continues to impact billions of people worldwide, and measurement of it remains an important focus of research. However, most Internet censorship measurements have focused solely on the IPv4 Internet infrastructure. Yet, more clients and servers are available over IPv6: According to Google, over a third of their users now have native IPv6 access. Given the slow-but-steady rate of IPv6 adoption, it is important to understand its impact on censorship. In this paper, we measure and analyze how censorship differs over IPv6 compared to the well-studied IPv4 censorship systems in use today. We perform a comprehensive global study of censorship across an array of commonly censored protocols, including HTTP, DNS, and TLS, on both IPv4 and IPv6, and compare the results. We find that there are several differences in how countries censor IPv6 traffic, both in terms of IPv6 resources, and in where and what blocklists or technologies are deployed on IPv6 networks. Many of these differences are not all-or-nothing: we find that most censors have some capacity to block in IPv6, but are less comprehensive or less reliable compared to their IPv4 censorship systems. Our results suggest that IPv6 offers new areas for censorship circumvention researchers to explore, providing potentially new ways to evade censors. As more users gain access to IPv6 addresses and networks, there will be a need for tools that take advantage of IPv6 techniques and infrastructure to bypass censorship.