Privacy-Preserving Correlated Heavy-Hitters Computation Over Outsourced Cloud

Heavy hitters mainly refers to the element that appears most frequently in the data stream, and heavy hitters computation is a core problem in data mining and network traffic monitoring, when the data stream has two-dimensional features, which we call correlated heavy hitters. However, traditional heavy hitters computation methods require direct access to the original data, which may lead to privacy leakage, especially in cloud computing environments, where the users often do not want cloud service providers to view their data directly. Therefore heavy hitters computation faces the challenge of privacy protection, in some two dimensional data streams, such as source and destination addresses, time and location, the risk of privacy leakage is more serious. To solve this problem, this paper mainly uses Yao's garbled circuit and secret sharing to construct a secure correlated heavy hitters computation scheme, and performs security analysis to prove that this scheme is secure under the semi-honest model, and finally analyzes the operation efficiency of this scheme through a large number of experiments.