Privacy-preserving weighted similarity query over encrypted healthcare data
University of New Brunswick
The development of smart applications in the healthcare area has aroused the exponential growth of healthcare data. Similarity range query, which is purposed to search for similar objects in a dataset with particular metrics, has been widely ap plied in a variety of practical applications, including the disease diagnosis scenario. As a result, the data owner of those healthcare data tends to outsource them to powerful cloud servers and the latter can provide query services for the doctors with similarity range query. However, for privacy concerns, the data owner may outsource the encrypted data instead of plaintexts to the cloud servers. Meanwhile, when using the data query service, sometimes query users may search the data based on their preference. To address the aforementioned issues, in this thesis, we propose WeightedSim, an efficient and privacy-preserving weighted similarity range query scheme for outsourced healthcare data. Specifically, we first develop an encrypted R-Tree index by utilizing the Symmetric Homomorphic Encryption (SHE) technique and then employ it to perform a weighted similarity range query under the two cloud servers model. We analyze the security of our scheme to be selectively secure when the SHE is semantically secure against CPA and also conduct extensive experiments to validate the scheme’s efficacy.