Privacy Preserving Nearest Neighbor Query Processing And Privacy Preserving Hamming Distance Query Processing On Encrypted Data For Cloud Computing

Cloud computing is a new computing paradigm after distributed computing,peer to peer computing,and grid computing.It's core idea is resource renting,application hosting,and services outsourcing.Cloud has a high flexibility and scalability,and its applications have become more and more popular.Nearest neighbor querying is one of the most fundamental operation.It has many applications in various fields such as pattern recognition,machine learning,and statistical classification.Hamming distance has been widely used in many application fields,such as near-duplicate detection,pattern recognition,virus detection,DNA sequence,and protein analysis,error correction,and document classification.With the advent of the cloud computing,outsourcing massive datasets to cloud servers has become more and more popular,but we can't fully trust the cloud for the potential threats.In order to ensure the security of the data,data owner usually encrypt the dataset before outsourcing it to the cloud.So it is urgent to work out fast and secure nearest neighbor and Hamming Distance search solutions under the encrypted data.We proposed secure searching protocols accordingly.Our main research work includes the following aspects:Firstly,we proposed an SNN solution for 1-D space and 2-D space.The basic idea is to convert the problem of finding the nearest neighbor to the problem of testing whether a query point is in a range and then further to the problem of testing whether two sets have common elements,and then solve the problem of testing set intersection in a privacy preserving manner.In order to achieve logarithmic query time,we proposed a range tree data structure.To achieve adaptive security under IND-CPA model,we propose a multi-homing Bloom filter for building the secure index and a single-homing Bloom filter for building trapdoors.The experiment results validate the efficacy and efficiency of the proposed scheme.Secondly,we proposed a privacy preserving hamming distance query processing scheme.First,we randomly generate an invertible matrix to transform the data into medium results.Second,we use the prefix encoding technique to encode the medium results.Third,we use the secure data structure IBF to store the encoded results and build an IBTree index to achieve sub-linear query time.As IBtree scheme is secure under IND-CKA model,our scheme is also secure under IND-CKA model.Our experimental results validate the efficiency of the proposed scheme.We summarized our work and proposed further work in the conclusion.