20th Annual Network & Distributed System Security Symposium
February 24-27, 2013, San Diego, CA
Open communication over the Internet poses a serious threat to countries with repressive regimes, leading them to develop and deploy censorship mechanisms within their networks. Unfortunately, existing censorship circumvention systems face difficulties in providing unobservable communication with their clients; this highly limits their availability as censors can easily block access to circumvention systems that make observable communication patterns. Moreover, the lack of unobservability may pose serious threats to their users. Recent research takes various approaches to tackle this problem, however they introduce new challenges, and the provided unobservability is breakable.
In this paper we propose an easy-to-deploy and unobservable censorship-resistant infrastructure, called FreeWave. FreeWave works by modulating a client’s Internet traffic into acoustic signals that are carried over VoIP connections. Such VoIP connections are targeted to a server, the FreeWave server, that extracts the tunneled traffic and proxies them to the uncensored Internet. The use of actual VoIP connections, as opposed to traffic morphing, allows FreeWave to relay its VoIP connections through oblivious VoIP nodes (e.g., Skype supernodes), hence keeping the FreeWave server(s) unobservable and unblockable. In addition, the use of end-to-end encryption, which is supported/mandated by most VoIP providers like Skype, prevents censors from distinguishing FreeWave’s VoIP connections from regular VoIP connections.
To utilize a VoIP connection’s throughput efficiently we design communications encoders tailored specifically for VoIP’s lossy channel. We prototype FreeWave over Skype, the most popular VoIP system. We show that FreeWave is able to reliably achieve communication throughputs that are sufficient for web browsing, even when clients are far distanced from the FreeWave server. We also validate FreeWave’s communication unobservability against traffic analysis and standard censorship techniques.