Outsourced Privacy-Preserving Random Decision Tree Algorithm Under Multiple Parties for Sensor-Cloud Integration

摘要

The emerging trends in cloud computing have facilitated the integration of existing technologies towards achieving new and innovative applications for the betterment of humans. Remote health monitoring, a bi-product of technology integration, assists in minimizing human mortality through continuous health monitoring using low-cost sensors. However, privacy and security concerns have become a bottleneck in this process. The secure multi-party computation (SMC)-based privacy-preserving data mining algorithm has emerged as a solution to this problem. However, traditional cryptography-based PPDM solutions are too inefficient and infeasible for analysis on large-scale datasets for data owners. Previous work on random decision trees (RDTs) shows that it is possible to generate equivalent and accurate models at substantially lower costs. In this paper, we focus on the outsourced privacy-preserving random decision tree (OPPRDT) algorithm for multiple parties. We outsource most of the protocol computation to the cloud and propose secure sub-protocols to protect users' data privacy. As a result, we show that our method can achieve similar results as the original RDT algorithm while also preserving the privacy of the data. We prove that there is a sub-linear relationship between the computational cost of the user side and the number of participating parties.