Perfectly-Secure Asynchronous MPC for General Adversaries (Extended Abstract)

摘要

We study perfectly-secure Multiparty Computation (MPC) in the asynchronous communication setting, tolerating a generalized non-threshold adversary, characterized by an adversary structure. Ashwin Kumar et al. (ACISP2002) presented a condition which is both necessary as well as sufficient for the existence of perfectly-secure MPC in this setting. However, we show that their protocol is flawed and present a new protocol (with the same necessity condition). Moreover, our protocol is conceptually simpler, and unlike their protocol, does not rely on monotone span programs (MSPs). As a sub-contribution, we also present an asynchronous Byzantine agreement protocol (tolerating a non-threshold adversary), which is used as a key component in our MPC protocol.