Paillier Cryptosystem Optimisations for Homomorphic Computation.

Abstract

Homomorphic encryptions can ensure privacy in systems operating with sensitive data. It also allows outsourcing the data processing without the need to disclose the information within. To keep good performance over the growing mass of data, the execution of homomorphic schemes has to be efficient. In this article, we focus on the optimization of the Paillier scheme. This scheme allows the addition of a constant or another ciphertext without decryption of the encrypted values. Since the exponentiation used in the encryption process is time-consuming, we have implemented noise and message pre-computation to avoid time-demanding operations. These adjustments significantly fasten the encryption process, especially using the noise pre-computation.