Jesper Buus Nielsen has accepted to deliver an Invited talk at SCN 2020!
Title: You Only Speak Once
We introduce the You Only Speak Once (YOSO) MPC model along with some basic feasibility results about the model.
Since the introduction of the Bitcoin protocol the cryptographic and distributed literatures have seen a flourish in protocol designs where the nodes speak at random unknown points in time, send a single message, and then never have to speak again, or at least not keep mutable state until the next random activation. The motivation for this design pattern is protection against distributed denial-of-service (DDoS) attacks. The dramatic setting imagined is one where whenever a machine sends a message to perform some role in a protocol, its IP address becomes known to the adversary, and it is then immediately killed by a DDoS attack. This prevents the machine from passing on state and/or ever sending a message again. Ensuring by design that the machine executing a given role acts at a random point in time and never has to act twice provides very high mitigation of such DDoS attacks. This YOSO design pattern is related to other common design patterns in distributed systems like serverless architectures.
We are interested in studying the YOSO design pattern from a cryptographic point of view. In particular, can we do secure multiparty computation (MPC) in the YOSO model? And if so, how securely and efficiently can we do it?
In this talk we will introduce a simple, abstract model called the YOSO model. The YOSO model is meant to abstract away a lot of irrelevant real world details and allow us to focus on the essential parts of the YOSO design pattern. It is meant as a low cost entry point into YOSO MPC research.
We then survey some recent results on YOSO MPC and show two new feasibility results.
We show that it is possible to do computationally secure YOSO MPC with guaranteed output delivery against corrupted minorities with a protocol which is relatively practical.
We show that it is possible to do information theoretically secure YOSO MPC with guaranteed output delivery against corrupted minorities with a protocol which is not practical.
We will also propose a number of interesting open problems in YOSO MPC.
The purpose of the talk is to try to draw more people into what we believe is an interesting up-and-coming research area of YOSO MPC.