Multiple Linear Regression Based on Stream Homomorphic Encryption Computing

Abstract

This study proposes a multiple linear regression architecture based on stream homomorphic encryption computing to analyze ciphertext for massive secure data computing. The proposed architecture contains three subsystems including terminal subsystem, data access subsystem, and data computing subsystem. The method used behind the presented architecture contains four stages which are data preprocessing stage, data access stage, data computing stage, and result processing stage. In the practical experiments, a case study of traffic information prediction was selected to evaluate the proposed system and method. The predicted traffic information was generated by the proposed method in accordance with the encrypted traffic information. Our experimental results showed that the proposed architecture can effectively and promptly obtain the predicted traffic information.