Publicly Verifiable Secure Computation and Applications

Secure computation has been extensively studied in the cryptography literature
aiming at obtaining more efficient protocols. However, even though state-of-theart protocols allow large number of parties to efficiently compute a function
without leaking their inputs, they do not allow for third parties (i.e. who do not
participate in the protocol) to verify that a given output was correctly obtained by the computing parties. This property is called public verifiability and has been
proven to be achievable, though with very inefficient protocols. The inherent
inefficiency of current approaches to publicly verifiable secure computation
precludes its adoption for a number of applications such as secure computation
on decentralized ledgers (e.g. blockchains), where third parties must be able to
independently validate all transactions in the ledger. In this project, we explore
new approaches for constructing concretely efficient publicly verifiable secure multiparty computation (MPC) protocols, understanding both the fundamental
limits of this class of protocols and developing techniques for constructing them. We will answer the following main questions:
1. What is the fastest way to produce a publicly verifiable proof that a
given output was obtained by a MPC protocol? What is the smallest size
for such a proof?
2. How can we construct efficient publicly verifiable MPC in the
preprocessing model optimizing overall complexity in the online phase?
3. How can we construct efficient publicly verifiable MPC with constant
rounds for deployment over high latency networks (e.g. the Internet)?
4. Can we leverage state channels (e.g. for off-chain micropayments) and
smart contracts to obtain better publicly verifiable MPC?

The main results of this project were: CRAFT, a modular construction of MPC with financial transactions with cheater identification and the novel notion of output independent abort, which serves as basis for realising privacy preserving smart contracts; Eagle, a framework for MPC-based privacy preserving smart contracts with confidential transactions; TARDIS, a model for dealing with time in the UC framework and UC secure constructions of time-lock puzzles.