Paper 2018/021

Regularly Lossy Functions and Their Applications

Yu Chen and Baodong Qin and Haiyang Xue

Abstract

In STOC 2008, Peikert and Waters introduced a powerful primitive called lossy trapdoor functions (LTFs). In a nutshell, LTFs are functions that behave in one of two modes. In the normal mode, functions are injective and invertible with a trapdoor. In the lossy mode, functions statistically lose information about their inputs. Moreover, the two modes are computationally indistinguishable. In this work, we put forward a relaxation of LTFs, namely, regularly lossy functions (RLFs). Compared to LTFs, the functions in the normal mode are not required to be efficiently invertible or even unnecessary to be injective. Instead, they could also be lossy, but in a regular manner. We also put forward richer abstractions of RLFs, namely all-but-one regularly lossy functions (ABO-RLFs) and one-time regularly lossy filters (OT-RLFs). We show that (ABO)-RLFs admit efficient constructions from both a variety of number-theoretic assumptions and hash proof system (HPS) for subset membership problems satisfying natural algebraic properties. Thanks to the relaxations on functionality, the constructions enjoy much compact key size and better computational efficiency than that of (ABO)-LTFs. We demonstrate the utility of RLFs and their extensions in the leakage-resilient cryptography. As a special case of RLFs, lossy functions imply leakage-resilient injective one-way functions with optimal leakage rate $1-o(1)$. ABO-RLFs (or OT-RLFs) immediately imply leakage-resilient one-time message authentication code (MAC) with optimal leakage rate $1-o(1)$. ABO-RLFs together with HPS give rise to leakage-resilient chosen-ciphertext (CCA) secure key encapsulation mechanisms (KEM) (this approach extends naturally to the identity-based setting). Combining the construction of ABO-RLFs from HPS, this gives the first leakage-resilient CCA-secure public-key encryption (PKE) with optimal leakage rate based solely on HPS, and thus goes beyond the barrier posed by Dodis et al. (Asiacrypt 2010).

Note: An extended abstract of this paper appears in CT-RSA 2018. This is the full version.