VeraGreg: A Framework for Verifiable Privacy-Preserving Data Aggregation

摘要

A lot of effort has been made to devise a scheme for verifiable and privacy-preserving outsourcing of arbitrary computations. However, such schemes rely on Fully Homo-morphic Encryption which is still far from practical. In our work, we instead focus solely on encryption schemes with single homomorphic operation, in particular addition. We define a rigorous framework that gives the data originator a possibility to check what values have been incorporated within provided homomorphic aggregate. We also propose a practical scheme that instantiates this framework and prove that it achieves Indistinguishability under Non-Adaptive Chosen Ciphertext Attack (IND-CCA1). The definition of our framework led us further to a straightforward modification of the security notions of Non-Malleability (NM) and Adaptive Chosen Ciphertext Attack (CCA2). Our modification aims at preventing trivial breach which is by principle unavoidable for plain homomorphic encryption. With our enhancement, the notions of security can serve as a novel security goal for any future verifiable homomorphic schemes.