Parallel Algorithm Of Fully Homomorphic Encryption Over Floating-Point Based On Spark

With the rapid development of computer technology and mobile Internet,a large amount of user data is uploaded and stored in cloud servers,including personal privacy data.The development of cloud computing brings convenience to people,as well as security problem,such as privacy preserving,which increased people's concerns,especially the frequent occurrence of cloud security incidents.Therefore,the security of cloud computing and privacy protection has become a hot topic gradually,and there is an increasing demand for searching and processing of the encrypted data.It has become an urgent problem to be solved that how to guarantee the confidentiality of data stored in the untrusted cloud server,especially the sensitive information of users while they are enjoying the convenience of cloud computing.One of the major ways to solve this problem is to utilize fully homomorphic encryption(FHE)algorithm to support operations on the encrypted data directly.However,it is difficult to apply them to reality because most FHE schemes only support limited data types for the time being,and,often with lower efficiency and bigger space occupation for ciphertext.In 2015,LFHE,a scheme that does not need any noise reduction mechanism,was proposed.LFHE has higher efficiency and lower ciphertext expansion rate,but soon was proved to be insecure.Based on the core design of LFHE,the thesis proposed a fully homomorphic encryption scheme FFHE that supports floating-point operation,which increases the data types supported by the algorithm and improves the security of the scheme.Then,the parallelization of FFHE encryption algorithm,based on Spark,was proposed for processing large-scale data in cloud computing environment.The security and performance of the parallel algorithm were analyzed in theory and experiments were conducted to demonstrate its practical performance.Finally,this thesis designed a variety of application models for cloud computing,which provide a sample for the application of FFHE scheme in reality.Theoretical analysis and experimental results show that FFHE has higher efficiency and security guarantee.And the overall speed-up ratio of the parallel algorithm can reach 3.9 in a 4-node 16-core cluster and encryption and calculation time on encrypted data can be reduced effectively.Thus FFHE can satisfy the encryption requirement of large-scale data in cloud environment.