Context
Anonymous communication channels allow parties to broadcast messages in distributed networks, while hiding their own identity. That is, everybody can see the message, but nobody knows who is the sender. Such channels are powerful for two main reasons. First, they allow parties to speak freely without fear of legal consequences or personal retaliation by oppressive regimes or entities. Second, they allow for building more robust distributed systems, as hiding the sender of certain messages can, for example, make denial-of-service attacks more difficult to carry out.
Anonymous communication channels should provide two main properties, namely liveness and anonymity. These properties ensure that a sender’s message is eventually seen by all and that their identity remains hidden, even when adversarial forces try to disturb the channel.
Open Research Questions
Despite the importance of such communication channels, we currently do not have fully satisfying realizations thereof. Approaches either suffer from high bandwidth/computational costs or provide unclear amounts of privacy.
Efficient Anonymous Communication Channels:
Anonymous communication systems have received a large amount of research interest over the years. A comprehensive survey on existing efforts was recently published and can serve as a good starting point on this topic. Even more recently Cho et al. showed that such channels can be constructed with provable security and reasonable bandwidth overheads, but unfortunately, their solution incurs prohibitively large computational costs, rendering their approach not scalable. An important open question is to construct more efficient channels, for example, ones that are faster than those constructed by Cho et al.
Efficient Anonymous Communication Channels via TEEs:
Trusted execution environments (TEEs) can be seen as trusted, private machines. TEEs allow for ensuring that computations are performed correctly and privately.
In a nutshell, TEEs allow for lifting distributed protocols that enjoy good privacy properties in the presence of passive adversarial observers, to protocols that are both private and secure in the presence of an actively interfering adversarial entity. TEEs shift the burden of achieving active security from expensive cryptographic techniques to the TEE hardware.
Recent works, such as ZipNet, have shown that TEEs can allow for significantly faster anonymous communication channels. The main drawback of their work is the weak liveness guarantees that it provides. If a single party from a set of ``communication servers'' is offline, then their entire system halts and no messages are delivered. Naturally, an open question is to harness the power of TEEs to design efficient anonymous communication channels provide liveness, even if some of the machines participating in the protocol stop responding or are outright malicious.
Ongoing Work at Flashbots:
We are currently actively investigating the practicality of anonymous communication channel designs that leverage TEEs. Towards this goal, we are looking at combining secret sharing based secure computation techniques with TEEs to obtain channels that provide both privacy and liveness.
