Security Against Related Randomness Attacks via Reconstructive Extractors

This paper revisits related randomness attacks against public key encryption schemes as introduced by Paterson, Schuldt and Sibborn (PKC 2014). We present a general transform achieving security for public key encryption in the related randomness setting using as input any secure public key encryption scheme in combination with an auxiliaryinput reconstructive extractor. Specifically, we achieve security in the function-vector model introduced by Paterson et al., obtaining the first constructions providing CCA security in this setting. We consider instantiations of our transform using the Goldreich-Levin extractor; these outperform the previous constructions in terms of public-key size and reduction tightness, as well as enjoying CCA security. Finally, we also point out that our approach leads to an elegant construction for Correlation Input Secure hash functions, which have proven to be a versatile tool in diverse areas of cryptography.