dc.description.abstract | The cloud space is occupied with multiple versions of the same data, including images, text files, et. Cetera. Though cloud service providers (CSPs) are provid- ing very cost effective storage space but combating the unprecedented growth of data requires ways to optimize the data and minimize storage costs. Therefore the demand for data deduplication arises, which enables the removal of extra copies of the same data and manages the storage space efficiently. In this thesis, we proposed two frameworks, SDIHE and Privacy Preserving Disease Prediction and Secure Data Deduplication of Health Data. These frameworks deal with secure data deduplication and maintaining the privacy of users� data. In the first framework, we propose secure data deduplication approach to identify the duplicateor near duplicate images without actually looking at the underlying content. It is based on the client-server model where the client encrypts the image prior to outsourcing it to the server, and the server maintains only the unique copies at its end. The client�s privacy is preserved in the entire process as the data is encrypted using homomorphic encryption. Further, the data integrity is checked both at the server-side while uploading and client-side during download. Experiments arecarried out to verify the proposed approach�s performance against a variety of potential attack scenarios, such as poison attacks, dictionary attacks, sidechannel attacks, and frequency analysis attacks. In the second framework, we propose a framework that predicts the diseases based on the user�s symptoms without com- romising the user�s rivacy. It also performs secure data deduplication on the disease prescription to inimize the storage requirement. Instead of giving pre-scriptions to every patient individually, nly a unique copy of the prescription is maintained at the CSP, and the access table is updated accordingly for the patients. The efficiency of the proposed framework is validated through the experiments. | |