SoK: Opportunities for Accelerating Multi - Party Computation via Trusted Hardware

Multi-Party Computation (MPC) allows a set of parties to evaluate a joint function without revealing their private inputs to each other. However, existing MPC protocols suffer from high computation and communication complexity, making them incapable of practical privacy-preserving applications. Unlike MPC, which relies on expensive cryptography, Trusted Execution Environments (TEEs) aim to efficiently protect the integrity and confidentiality of run-time code and data from the hardware level. Despite this, TEE is not a panacea due to its vulnerability to various attacks to deteriorate their confidentiality. Therefore, integrating MPC protocols with TEE emerges as a promising approach to achieving both high performance and security. In this paper, we surveyed existing works of integrating TEE into MPC to build a privacy-preserving framework with optimized communication and computing overhead. These integrated MPC-TEE protocols vary from different trust assumptions of TEEs' confidentiality, by which they can be categorized into four classes: TEE with complete confidentiality, TEE with access pattern leakage, TEE with no confidentiality or integrity, and hybrid models of the above three cases. By analyzing these works, we summarized the focal points of integrating two approaches under different trust levels and provided paradigm cases for each level. We also figured out several challenges in integrating TEE and MPC protocols and suggested several future research directions that are worthy of further study.