Accelerating ElGamal Partial Homomorphic Encryption with GPU Platform for Industrial Internet of Things

Abstract

Industrial Internet of Things (IIoT) is a paradigm shifting technology used for enhancing the manufacturing process. Under this technology, cloud server is widely used to store sensor data collected through various sensor nodes. The stored data is being used for various analysis and computation, including machine learning and statistics, in hope of finding useful inferences for industrial purpose. However, some of these sensor data and analysis results are sensitive, which should be handled properly to protect the user privacy. In addition, the cloud server environment may not be fully trusted, due to potential cyber security attack by adversaries. Encryption is one of the most straightforward way to protect user privacy, but encrypted sensor data prohibits the cloud server to perform any further analysis. Homomorphic encryption is useful to provide encryption to sensor data, yet allow the third party (cloud server) to perform computation on the encrypted data. However, homomorphic encryption algorithm is usually complex and require a lot of computational effort. In this paper, we propose implementation technique to accelerate the ElGamal partial homomorphic encryption in GPU platform. This implementation allows homomorphic multiplication to be performed on cloud server for IIoT applications at high performance, yet able to protect the user privacy.