A Secure Cloud Framework for Big Data Analytics

Abstract

ABSTRACT : Big data analytics and its security have become important as more organizations use the cloud to host data. Cloud computing is one of the leading technologies that organizations rely on and has been widely used. Hosting big data on the cloud presents many challenges for organizations as it needs fast analytics and secure hosting. This thesis presents a big data framework that provides fast analytics as well as secure cloud hosting. The proposed framework uses distributed clustering techniques to improve big data analytics and hybrid encryption methods to enhance data security on the cloud. Distributed clustering divides big data into several subsets. In this method, numerous virtual machines process big data simultaneously to reduce processing time. The K-Means and Fuzzy C-Means clustering algorithms are used as two instances of data clustering using different sizes of datasets. Moreover, the principal component analysis (PCA) method is added to further reduce the clustering time Furthermore, to secure data on the cloud, the proposed framework uses a hybrid encryption solution based on Elliptical Curve Cryptogra-phy (ECC) as a lightweight encryption method and Fully Homomorphic Encryption (FHE). Homomorphic Encryption is a sophisticated cryptography system that can maintain data privacy in all cases when the data is at rest, in processing, and even in computing. The user or cloud provider can do all necessary computations without the need to decrypt the data. Using the mentioned hybrid encryption method can enhance data security while, at the same time, data analytics can be calculated on ciphertext data. The results provided in the proposed framework are compared with the centralized computation model to ap-prove its efficiency and reliability. The results show that the developed framework using the mentioned techniques can improve the performance of big data analytics while, at the same time, data is protected and privacy is preserved.