Securely computing threshold variants of signature schemes (and more!)
Eysa Lee · 2023
This thesis looks at two lines of study aiming at bringing privacy-enhancing technologies closer to deployment: using techniques from multiparty computation (MPC) to distribute digital signature algorithms; and improving MPC protocols tailored for models of computations more relevant for applications such as privacy-preserving machine learning.Much of cryptography makes use of private, hidden information, and security of the scheme falls apart if a bad actor gains access to the secret information. Threshold cryptography focuses on distributing cryptographic algorithms to require participation from a threshold number of parties (each holding some partial secret information) in order to produce an output. In this thesis, we highlight work done producing threshold variants of two signature schemes, the Elliptic Curve Digital Signature Algorithm (ECDSA) and BBS+. For these tasks, we make use of a novel secure multiplication subprotocol based off of the cryptographic primitive Oblivious Transfer. We are able to take advantage of the simple structure of each signing algorithm and the public verifiability inherent to signature schemes to obtain malicious security with little overhead to the semihonest case. Finally, we look at cryptographic techniques compatible for use in broader security and privacy applications. In particular, we cover a work aimed at improving efficiency generic MPC by enabling amortization of computing repeated subcircuits.--Author's abstract