A novel searchable encryption scheme to reduce the access pattern leakage

Abstract

Searchable symmetric encryption (SSE) is a technology that enables clients to outsource sensitive data in the cloud. The vast usage of cloud and rapid growth in file sharing over the cloud has forced researchers into devising newer searchable encryption (SE) systems that are more secure and robust. This paper explores the effects of leakage of access patterns to an adversary. It is observed that in most of the existing systems, a successful statistical attack can help the adversary learn the access pattern, provide him with the required document and reveal the underlying data. This paper addresses these issues by presenting a novel SE system which hides the search pattern and access pattern. In this paper, we propose a SSE system that uses forward index while lifting the one-time search limitation of the previous solutions. Our system uses a trapdoor obfuscation technique to deceive the 3rd party adversary who is considered as an active entity other than client and server, observing the information being exchanged during the active session of client and the cloud service provider. Compared with the existing techniques, this solution uses keyword analysis to extract limited words from the documents based on the frequency of occurrence. To provide stronger privacy, an encoding technique is used which aims to hide the access pattern from the adversary. In order to eliminate trapdoor linkability, a probabilistic trapdoor generation process is presented. The test results demonstrate that our system is suitable for practical usage with minimal overhead and is more efficient as compared to the state of the art.