Research On Multi-party Statistical Computation Based On Functional Encryption

Multi-party statistical computation has a wide range of applications in real life,but the statistical computation usually involves the sensitive data of multiple participants.If the privacy of sensitive data can't be guaranteed,the participants are unwilling to share their sensitive data for normal statistical computation.In above circumstances,privacy-preserving multi-party statistical computation has become an emerging need.The main work of this thesis is as follows:1.A multi-input functional encryption scheme for the privacy protection in multi-party statistical computation is proposed.Based on the design idea of the classical functional encryption scheme,the multi-input functional encryption scheme is constructed,with the technology of the multi-key fully homomorphic encryption,the two-outcome attribute-based encryption,and the garble circuit.Firstly,the two-outcome attribute-based encryption is used to encode the input label set for garble circuit.And then,the multi-key fully homomorphic encryption is used to compute over ciphertext which is encrypted by different keys,and the garble circuit is used to garble the decryption algorithm of multi-key fully homomorphic encryption.In this scheme,each participant has a unique key to encrypt the data,and the evaluator will execute computations for the encrypted data,such that this scheme can protect the data privacy of each participant effectively.2.A multi-party statistical computation scheme is designed to remove tedious protocol interaction in existing scheme.Multi-input functional encryption and STTP(Semi-trusted Third Party)computation model are used in this statistical computation scheme.Different participants use different public keys to encrypt their private data,then send the encrypted data to the server.The server performs statistical computation on the encrypted data,which means the server can't access the participant's original data during the computation process.This scheme protects the participant's data privacy without interaction under complex protocol.3.A statistical computation scheme based on multi-input functional encryption and tree topology is developed for massive data application scenarios.This scheme decomposes the calculation of the Statistics into a number of intermediate statistical results,and uses the aggregate tree to aggregate the intermediate statistical results.The statistical server performs statistical computation on the encrypted intermediate statistics.This scheme utilizes the computational power of the statistical nodes to reduce the time consumption of data aggregation and protect the data privacy in the statistical computing process.Two statistical computation schemes are designed according to different scenarios in this paper,and the security analysis demonstrated that our schemes are effective in protecting privacy of participants' data.