CLASSIFICATION OF SECURE ENCRYPTED RELATIONALDATA IN CLOUD COMPUTING

S.Narayanan, Naushin Ghani.V, Sangeetha.P, Manimegalai.K, Vishali.P

Abstract: Due to the increasing popularity of cloud computing, organisations have the choice to outsource their large encrypted data content along as well as data mining operations to cloud the environment. Outsourcing data to such a third party cloud environment can compromise the data security as cloud operations and data mining tasks cannot carry out computations without decrypting the data. Hence, already present privacy-preserving data mining techniques are not efficient to address the security and confidentiality problems. In the base paper, a k-NN classification algorithm over secure data under a semi-honest model was developed using a Paillier cryptosystem for public key encryption. The usage of public key cryptosystems has security issues during data transfer in the cloud. In this proposed work, we focus on solving the k-NN problem over secure encrypted data by proposing a privacy preserving k-nearest neighbour classification on encrypted information in the cloud using private key for encryption and decryption based on the symmetric AES cryptographic algorithm under the secure multiparty computations for creating a complete homomorphic encryption (CHE) scheme which results in the reduction of space requirement and processing time. Also, we aim to apply the same PPk-NN classification over encrypted images. The proposed protocol hides the input query and data access patterns of the users and also preserves the confidentiality of text and image data.Finally, we present a practical analysis of the efficiency and security performance of our proposed protocol for application in a Life insurance firm where the clients are classified according to their risk-level.

Keywords: Data Mining, PPk-NN, Semi-Honest Model , Individual Key, Symmetric Homomorphic Encryption, AES Algorithm,CHE, Less Space and Time.

DOI: https://doi.org/10.15623/ijret.2016.0503034