Similarity based Encrypted Data Search in Cloud Computing

Abstract

Encrypting con dential data before outsourced to cloud storage ensures data security and privacy but encryption hides data information making CSP unable to perform standard lookup queries for searching. This is the problem we addressed in our research to enable searching over outsourced encrypted data. Existing schemes for searching over encrypted data are based on trapdoors and utilizing locally stored indexes which limits searching capabilities to a limited pre-de ned trapdoors. Moreover, existing schemes support exact keyword matching. In this thesis we proposed a similarity based searching over outsourced encrypted data while ensuring end-to-end privacy. Unlike existing approaches, our scheme enables subscribers to de ne their own queries with arbitrary number of keywords. We proposed a private matching algorithm which ensures that cloud computing service performs term matching without revealing anything information about user query and con dential data by utilizing homomorphic encryption. We proposed a novel idea for reducing communication cost overhead and our results shows gain in communication cost reduction over 95%. We have implemented the scheme and our results have demonstrated that search queries with 2 to 10 keywords only cost 0.00001 to 0.00004 $ per 1000 similar requests.