MCMix: Anonymous Messaging via Secure Multiparty Computation
summary
MCMix implements anonymous messaging by splitting dialing and conversation into MPC routines, implemented on Sharemind and proven to provide simulation-based security against a global adversary. It uses oblivious sorting and identity-based key agreement, supports pairwise dead-drop hashing, and keeps client work to symmetric crypto. Benchmarks show about 100,000 users served with ~1–2 minute latency (half a million at ~4–5 minutes) and server bandwidths roughly 150–300 Mbps, and the prototype server-side implementation meets these targets.
AI-written summary · about
chapters (35)
- 0:00Intro
- 0:37Anonymous communications
- 2:00Anonymous messaging against global adversaries
- 3:26Our contributions
- 4:29Cryptographic Model: Client-Server Secure MPC
- 5:55Simulation-based Secure MPC
- 6:12Modeling Anonymous Messaging
- 6:53Ideal Dialing Functionality
- 7:19Ideal Conversation Functionality
- 7:38An ideal standard for anonymous messaging
- 8:12The design challenge
- 8:30MCMix architecture
- 8:58How MCMix works - General Idea
- 9:31Identity-based Key Agreement (ID-KA)
- 10:24Registration phase (1)
- 11:14Encoding and secret sharing
- 11:32grouping and forwarding
- 12:11Decryption and reconstruction
- 12:22Secure realization of 2 protocols
- 12:42How Dialing works - an example (2)
- 15:11Dead drop calculation
- 16:05The Conversation protocol - MPC phase
- 16:56About dead drops
- 17:09Simulation-based Security of MCMix
- 18:09Benchmarks: Dialing running time
- 18:46Benchmarks: Conversation running time
- 19:06varying message size
- 19:26Benchmarks: Network bandwidth consumption
- 19:42Estimated client communication cost
- 20:29THigh level comparison to other systems
- 21:32Ongoing work - Android prototype
- 21:50Parallelizing the Conversation protocol
- 23:11Performance of the parallelized Conversation protocol
- 23:34Future work
- 24:23MCMix: Anonymous Messaging via Secure Multiparty Computation
related talks
01h 31mIntroduction to Secure Multi-Party Computation
06mNetwork-level Anonymity and Transaction Privacy
34mPrivacy Preserving Aggregation of Ad Conversions using MPC
01h 08mSecure Computation 1
58mSecure Multiparty Computation I
23mRecent progress in MPC-in-the-Head protocols
21mFrom Usability to Secure Computing and Back Again
40mApplied MPC