Research Topics for Mozilla Research Grants

Mozilla has recently put out an RFP for research proposals related to the integration of Tor in Firefox, specifically Tor scalability and performance:

"Mozilla has an interest in potentially integrating more of Tor into Firefox, for the purposes of providing a Super Private Browsing (SPB) mode for our users. Tor offers privacy and anonymity on the Web, features which are sorely needed in the modern era of mass surveillance, tracking and fingerprinting. However, enabling a large number of additional users to make use of the Tor network requires solving for inefficiencies currently present in Tor so as to make the protocol optimal to deploy at scale. Academic research is just getting started with regards to investigating alternative protocol architectures and route selection protocols, such as Tor-over-QUIC, employing DTLS, and Walking Onions. What alternative protocol architectures and route selection protocols would offer acceptable gains in Tor performance? And would they preserve Tor properties? Is it truly possible to deploy Tor at scale? And what would the full integration of Tor and Firefox look like?"

Tor has many lists of research ideas that need to be consolidated. With respect with this RFP, the most useful place to start is the performance section of our open research problems blog post. That section discusses:

• congestion control
• datagram Tor, and
• load balancing problems and ideas.

To help evaluate radical changes to Tor's reliability and congestion control model, we need frameworks to accurately measure side channels in Tor. Such frameworks are important to allow us to compare the security of various ways of reducing congestion on the Tor network via datagram designs or explicit congestion notification in current Tor.

Additionally, the exit censorship problem is also a major usability barrier to using Tor, as is denial of service, and research into cryptographic rate limiting systems based on newer anonymous credentials would also be in-scope.

Finally, with walking onions, we could drastically lower Tor's consensus interval to provide faster load balancing feedback. What are the consequences of this in terms of additional traffic to relays? How much benefit do we get for this? See the Load Balancing subsection of the Network Performance research topics for related load balancing ideas that could benefit from a faster consensus interval.