Cloud computing is a boon for both business and private use, but data security concerns slow its adoption. Fully homomorphic encryption (FHE) offers the means by which the cloud computing can be performed on encrypted data, obviating the data security concerns. FHE is not without its cost, as FHE operations take orders of magnitude more processing time and memory than the same operations on unencrypted data. Cloud computing can be leveraged to reduce the time taken by bringing to bear parallel processing. This paper presents an implementation of a processing dispatcher which takes an iterative set of operations on FHE encrypted data and splits them between a number of processing engines. A private cloud was implemented to support the processing engines. The processing time was measured with 1, 2, 4, and 8 processing engines. The time taken to perform the calculations with the four levels of parallelization, as well as the amount of time used in data transfers are presented. In addition, the time the computation servers spent in each of addition, subtraction, multiplication, and division are laid out. An analysis of the time gained by parallel processing is presented. The experimental results shows that the proposed parallel processing of Gentry's encryption improves the performance better than the computations on a single node. This research provides the following contributions. A private cloud was built to support parallel processing of homomorphic encryption in the cloud. A client-server model was created to evaluate cloud computing of the Gentry's encryption algorithm. A distributed algorithm was developed to support parallel processing of the Gentry's algorithm for evaluation on the cloud. An experiment was setup for the evaluation of the Gentry's algorithm, and the results of the evaluation show that the distributed algorithm can be used to speed up the processing of the Gentry's algorithm with cloud computing.
展开▼
