Abstract:
As technology is improving and changing rapidly, cloud security has become a challenging task. Consequently, there is a need for more powerful and robust techniques to secure the cloud. Meanwhile, due to the huge size of the provided big data on the cloud, other techniques and methods should be utilized to improve big data analytics and processing. The paper aims to provide a framework for secure and efficient processing and analysis of big data using a double layer of security that is based on Elliptical Curve Cryptography (ECC) and Fully Homomorphic Encryption (FHE). Additionally, a distributed model has been defined to partition big data into smaller data sizes processed by different numbers of virtual CPUs. In the defined distributed model, many virtual machines process different partitions of data parallelly and simultaneously to speed up the processing time of data. KMeans clustering algorithm is used in three datasets as an instance of data analytics to test the suggested framework. Furthermore, the produced results are compared with a centralized-based model to assess the productivity and efficiency of the distributed model. Besides, the principal component analysis (PCA) is applied to the used clustering algorithm to diminish the required clustering time by the distributed model. The results indicate that the clustering time can be reduced by up to 91%, and even with 18% more reduction in the execution time using the distributed model. The recommended solution can improve the effectiveness of big data analytics while guaranteeing the security of such data.
