Title: CLOAK: Enabling Confidential Smart Contract With Multi-Party Transactions

Abstract: In recent years, as blockchain adoption has been expanding across a wide range of domains, e.g., supply chain and digital asset, smart contracts' confidentiality has now become a fundamental demand for practical applications. However, while new privacy protection techniques are emerging, how existing ones can best fit development settings is understudied. State-of-art solutions lack architectural support of programming interfaces thus can hardly reach general developers. This paper proposes CLOAK, a pluggable and configurable framework for developing and deploying confidential smart contracts. The key capability of CLOAK is to allow developers to develop and deploy practical solutions to Multi-party Transaction (MPT) problems, i.e., to verifiably transact with secret parameters and states owned by different parties by simply specifying it. To this end, CLOAK allows users to specify privacy invariants in a declarative way, automatically generate runtime with enforced privacy, and deploy it to enable the MPT on existing platforms. Additionally, we identify the pitfalls and treats for achieving MPT, e.g., achieving public verifiability and resisting byzantine adversaries with minimal blockchain interaction. In our evaluation of both examples and real-world applications, developers manage to deploy business services on blockchain concisely by only developing CLOAK smart contracts, whose size is less than 13.5% of the deployed ones. Moreover, while previous works require at least $O(n)$ transactions to secure an MPT, CLOAK requires only 2 transactions and reduces gas cost by 29% on average. We believe that the insights learned from CLOAK pave the way for public verifiable thus reusable general-purpose multi-party computation achieved by harmonizing TEE and blockchain.