Paper 2022/1119

PESCA: A Privacy-Enhancing Smart-Contract Architecture

Abstract

Public blockchains are state machines replicated via distributed consensus protocols. Information on blockchains is public by default---marking privacy as one of the key challenges. We identify two shortcomings of existing approaches to building blockchains for general privacy-preserving applications, namely (1) the reliance on external trust assumptions and (2) the dependency on execution environments (on-chain, off-chain, zero-knowledge, etc.) with heterogeneous programming frameworks. Towards solving these problems, we propose PESCA---a privacy-enhancing smart contract architecture. PESCA utilizes generic building blocks such as threshold fully-homomorphic encryption (FHE), distributed key generation (DKG), dynamic proactive secrete sharing (DPSS), Byzantine-fault-tolerant (BFT) consensus, and universal succinct non-interactive zero-knowledge proofs (zk-SNARKs). First, we formalize the problem of replicating state machines augmented with threshold decryption protocols and discuss how existing BFT consensus protocols can be adapted to this setting. We describe how to instantiate a blockchain with a fixed FHE public key and have FHE-encrypted chain states programmatically decrypted via consensus. Next, we describe a smart-contract framework for engineering privacy-preserving applications, where programs are expressed---in a unified manner---between four types of computation: transparent on-chain, confidential (FHE) on-chain, user off-chain, and zero-knowledge off-chain. Lastly, to showcase the generality and expressiveness of PESCA, we provide two simple application designs for constant function market makers (CFMMs) and first-price sealed-bid auctions (FPSBAs), both with maximal privacy guarantees.