Adversarial analysis of internet censorship systems

Thesis

Many countries perform internet censorship in an attempt to control the information and material that its population can access. With no standard way that this is performed, most countries build a bespoke platform to carry out their aims. Information targeted for censorship is set at the state level, and then this is enforced by technical controls implemented either directly by the national government or by Internet Service Providers.

The details of these systems are rarely made public. As they each target information on the internet in different ways it can be difficult to gather details on what exactly they block, or how exactly they do it. This lack of knowledge can lead to situations where we experience unknown behaviour on the internet.

Lack of understanding can lead to unpredictable behaviour on the internet. Network traffic can be interfered with in ways that wasn't intended. In some cases this can happen in situations where neither the sender or receiver are under the direct jurisdiction of the censorship regime affecting them.

Unlike surveillance which can be passive, for censorship an observable action must take place. This gives us an opportunity to study these actions. We can observe when network connections are reset, packets are dropped, or incorrect results are given to queries. These actions allow us to study how such systems are implemented, and the nature of the content that they're blocking.

In this work we focus on the technical implementation of these internet censorship systems. How have they been implemented, and how do they go about blocking unwanted content? We initially focus on the Great Firewall of China, but eventually branch out to look at other internet censorship systems.

We also look at censorship avoidance technologies. Users in censorious regimes turn to technologies such as Tor or VPNs in an attempt to access information they would not otherwise be able to access. We explore how users use these technologies to access censored information, and whether the use of these technologies themselves can be correlated with internet censorship.

Authors: Farnan, OJ

• DOI: 10.5287/ora-kompqwmbj
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: censorship; privacy; security; internet
• Subjects: Censorship; Security; Privacy; Internet