A secure density peaks clustering algorithm on cloud computing

Abstract

Cloud computing provides users with the convenience of data outsourcing computing at risk of privacy leakage, and clustering algorithms have high computational overhead when dealing with large datasets. Aiming at the above problems, this paper presents a security density peak clustering algorithm based on grid in hybrid cloud environment. First, the client uses the homomorphic encryption method to build the encrypted objects with user datasets. Second, the client uploads the encrypted objects to the cloud servers to implement the security protocols proposed in this paper. Finally, the cloud servers return the perturbation clustering results to the client to eliminate the disturbance. In the proposed scheme, only encryption and removing perturbation are performed on the client, ensuring that the client has lower computational complexity. Security analysis and experimental results show that the scheme proposed in this paper can improve the efficiency and accuracy of clustering algorithm under the premise of protecting user privacy.