Differential Privacy Enhanced Dynamic Searchable Symmetric Encryption for Cloud Environments.

Peiyi Tu,Xingjun Wang

DOI: https://doi.org/10.1007/978-981-97-0945-8_22

2024-01-01

Abstract:In recent years, the Database-as-a-Service (DAS) model has become increasingly popular for cost-effective data outsourcing to cloud service providers. To ensure data security and functionality, Searchable Encryption (SE) has been introduced. However, many existing SE schemes unintentionally leak access patterns, posing significant privacy risks. While solutions like Oblivious RAM (ORAM) and Fully Homomorphic Encryption (FHE) can mitigate this, they are computationally intensive, limiting their practicality for large-scale databases. In this paper, we design a dynamic and efficient SE scheme called DPDSE that exploits differential privacy obfuscation of access patterns. Specifically, we obfuscate the ciphertext by deploying Laplace and Randomized Response mechanism. DPDSE strikes a balance between privacy and storage costs, while maintaining compatibility with any SE scheme. We give a formal mathematical proof of the security of DPDSE and conduct experiments on real datasets. The results show that DPDSE is significantly more secure than traditional SE schemes at a storage cost of up to 2.5%.