Outsourcing the Computation of Plaintext Encryption for Homomorphic Encryption

Homomorphic encryption is a promising technology for private enhanced computing. It provides a data-centric privacy-preserving method that allows computing on encrypted data by a third proxy. However, lightweight terminals lack sufficient computing resources to perform a large amount of encryption operations in stream data applications, e.g., smart devices in blockchain network, sensors and cameras in urban intelligent transport. This paper presents a blinded encryption paradigm that outsources most computations to a third proxy, e.g., cloud computing and edge computing. It helps to eliminate the encryption overhead for lightweight terminals while preserving data privacy. The proposed paradigm is composed of plaintext blinding, blinded encryption, ciphertext recovery and one round of communication. Lightweight terminals only execute efficient plaintext blinding and ciphertext recovery to guarantee data privacy. The expensive computing for plaintext encryption is delegated to a third proxy through blinded encryption. In addition, the key analysis to achieve time consuming improvement is to overlap the communication latency with computation time. Furthermore, the security analysis demonstrates the outsourced encryption is secure under the assumption of secure blinding and recovery algorithms. Finally, the experiment shows our method outperforms the original version by 2x improvement from the prospective of resource consumption.