Secure KNN Computation on Cloud

Cloud computing has emerged as a trend of outsourcing database and query services to a powerful cloud to ease local storage and computing pressure. However, private data storage and computation in the cloud come with a risk of losing the privacy and confidentiality of the data. Thus, sensitive applications running on the cloud require to be encrypted before storing it in the cloud. Additionally, to run some data mining algorithms, viz., k-NN requires the data to be in the encrypted domain computation-friendly ciphertext form. Thus, data are encrypted using a searchable encryption scheme before outsourcing it into the cloud. Asymmetric scalar-product-preserving encryption (ASPE) scheme provides secure k-nearest neighbors computation that is designed to provide both Data Privacy and Query Privacy. However, this scheme assumed that the query users were trusted entities. Enhancements to this work further showed that trusted query user assumption is no longer necessary if we use the Paillier cryptosystem. In this work, we have shown that even if we do not use the Paillier cryptosystem, query privacy in the cryptosystem can be achieved using the ASPE technique alone. Using ASPE for query encryption reduces the query encryption time, further improving the practicality of the encryption scheme.