Paper 2018/524

New Smooth Projective Hashing For Oblivious Transfer

Bing Zeng

Abstract

Oblivious transfer is an important tool against malicious cloud server providers. Halevi-Kalai OT, which is based on smooth projective hash(SPH), is a famous and the most efficient framework for $1$-out-of-$2$ oblivious transfer ($\mbox{OT}^{2}_{1}$) against malicious adversaries in plain model. A natural question however, which so far has not been answered, is whether its security level can be improved, i.e., whether it can be made fully-simulatable. In this paper, we press a new SPH variant, which enables a positive answer to above question. In more details, it even makes fully-simulatable $\mbox{OT}^{n}_{t}$ ($n,t\in \mathbb{N}$ and $n>t$) possible. We instantiate this new SPH variant under not only the decisional Diffie-Hellman assumption, the decisional $N$-th residuosity assumption and the decisional quadratic residuosity assumption as currently existing SPH constructions, but also the learning with errors (LWE) problem. Before this paper, there is a folklore that it is technically difficult to instantiate SPH under the lattice assumption (e.g., LWE). Considering quantum adversaries in the future, lattice-based SPH makes important sense.