Paper 2016/818

Composable Adaptive Secure Protocols without Setup under Polytime Assumptions

Carmit Hazay and Muthuramakrishnan Venkitasubramaniam

Abstract

All previous constructions of general multiparty computation protocols that are secure against adaptive corruptions in the concurrent setting either require some form of setup or non-standard assumptions. In this paper we provide the first general construction of secure multi-party computation protocol without any setup that guarantees composable security in the presence of an adaptive adversary based on standard polynomial-time assumptions. We prove security under the notion of ``UC with super-polynomial helpers'' introduced by Canetti et al. (FOCS 2010), which is closed under universal composition and implies ``super-polynomial-time simulation''. Moreover, our construction relies on the underlying cryptographic primitives in a black-box manner. Next, we revisit the zero-one law for two-party secure functions evaluation initiated by the work of Maji, Prabhakaran and Rosulek (CRYPTO 2010). According to this law, every two-party functionality is either trivial (meaning, such functionalities can be reduced to any other functionality) or complete (meaning, any other functionality can be reduced to these functionalities) in the Universal Composability (UC) framework. As our second contribution, assuming the existence of a simulatable public-key encryption scheme, we establish a zero-one law in the adaptive setting. Our result implies that every two-party non-reactive functionality is either trivial or complete in the UC framework in the presence of adaptive, malicious adversaries.