Robust and Confidential Multi-Tenant Keyword Retrieval in Cloud Storage

Abstract

The project addresses the critical security challenges of cloud storage by introducing a secure and efficient mechanism for keyword search over encrypted data, preserving user privacy without sacrificing retrieval performance. At its core is a multi-client architecture powered by Distributed Point Functions (DPFs), enabling keyword queries on encrypted datasets while concealing the query contents Keyword indexes are compactly stored using a Garbled Bloom Filter alongside a Cuckoo Filter employing cuckoo hashing. This design, combined with segmenting techniques to parallelize across multiple threads, significantly reduces computation time and network overhead—especially for conjunctive multi-keyword searches . To bolster security and access control, the system employs dual-layer encryption, leveraging AES for symmetric protection and ECC for asymmetric security, applied uniformly to user data and index structures. Integrity is enforced through Wegman–Carter message authentication codes and set-constrained pseudorandom functions, guarding against tampering and client collusion . Additionally, data compression and intelligent caching reduce storage needs and search latency, while a working prototype underlines the design's real-world efficiency through comprehensive performance evaluation