An NTRU-based Homomorphic Encrypted Data Analysis System

In the current data processing systems, there exist widespread threats to the privacy of data providers. To solve the problems in privacy protection in data processing, this paper proposes a homomorphic encryption data analysis system-MeSec, which is based on NTRU cryptosystem. Firstly, MeSec constructs a new type of password-based authorized key exchange protocol (PAKE). Secondly, MeSec adopts a more efficient NTRU-based homomorphic encryption system. And respectively designs a federated learning scheme for encrypted distributed training models and a privacy machine learning prediction scheme whit the encrypted private data as inputs; finally, an effective multiuser agent NTRU-based re-encryption module is designed to realize safe and convenient data intercommunication and sharing between users, which completely breaks the barriers of data islands. This system can be widely used in various fields for data sharing, analysis, and processing.