Elisaweta Masserova

Distributed Cryptography as a Service Degree Type: CS
Advisor(s): Bryan Parno, Vipul Goyal
Graduated: September 2024

Abstract:

Today's world is undeniably data-driven. The explosion of the Internet has generated vast volumes of data, and the advent of machine learning has unlocked captivating applications that thrive on this data. In such a world, it is evident that the ability to store, transmit, and process data securely is paramount. Distributing trust is one of the fundamental cryptographic principles that enable such security, and it is at the core of key cryptographic tools such as multi-party computation (MPC) and randomness generation. As the demand for secure and reliable cryptographic solutions grows, there is increasing interest in offering distributed protocols as a service.

Such services are typically expected to run continuously for long periods of time, requiring significant resource commitments from all participating parties. One approach to mitigate this issue is to design distributed cryptographic protocols that are stateless. In stateless protocols, parties are easily replaceable, and can contribute to the execution by participating only for a short time, without committing to a long-term computation.

In this work, we study stateless multi-party computation. We start by introducing a stateless MPC protocol which does not require parties to be online at the same time and requires no interaction between the participants. We construct this protocol in the blockchain model and under the assumption of what we call Conditional Storage and Retrieval (CSaR) systems. In our next step, we eliminate the CSaR requirement and design stateless MPC without relying on this assumption. More concretely, we focus on the recently introduced You Only Speak Once (YOSO) paradigm. In this model participating parties are allowed to send only a single message; i.e., they speak only once. We improve the state of the art in YOSO MPC by designing a protocol with better communication complexity than the currently known solutions. Then, we focus on improving the efficiency of special-purpose YOSO MPC. Specifically, we consider the task of distributed randomness generation, and design a suite of protocols, each balancing different trade-offs in terms of underlying assumptions, efficiency, and corruption threshold. We develop these protocols in the model of YOSO with worst-case corruptions (YOSO^WCC), which is even stronger than the original YOSO MPC.

Thesis Committee:
Bryan Parno (Co-Chair)
Vipul Goyal (Co-Chair, Carnegie Mellon University/NTT Research)
Eliane Shi
Antigoni Polychroniadou (J.P. Morgan AI Research)
Tal Rabin (Amazon Web Services / University of Pennsylvania)

Srinivasan Seshan, Head, Computer Science Department
Martial Hebert, Dean, School of Computer Science

Keywords:
Secure Multi-Party Computation, Distributed Systems, Blockchains