Homomorphic Encryption-Based Secure Data Retrieval in Cloud Storage

Abstract

Homomorphic encryption (HE), enabling computations on encrypted data without decryption, is a compelling approach for secure data retrieval in cloud storage. However, performance overhead and query inefficiencies have limited its widespread adoption. In this work, we propose a hybrid homomorphic encryption scheme for secure data retrieval in cloud storage systems, combining the strengths of multiplicative homomorphic encryption for bulk operations and fully homomorphic encryption (FHE) for fine-grained query processing. This hybrid model reduces encryption depth complexity—making retrieval operations scalable even over large datasets—while ensuring provable semantic security.

Recent advances such as a hybrid privacy information retrieval model introduce this dual-layer HE strategy to improve efficiency for large databases. Techniques include using assignment methods to reduce HE operations for serialized searches and employing FHE only for residual simple operations. These enhancements decouple FHE complexity from database size, improving responsiveness. Complemented by caching optimizations in database systems—such as singular caching (Silca) and radix-based additive caching (Rache)—HE retrieval latency is further reduced. 

Our implementation integrates the hybrid model into a cloud storage prototype supporting keyword-based search and aggregation queries. Experimental results demonstrate significant reductions in retrieval latency compared to traditional FHE-only approaches, while preserving strong confidentiality and query privacy. The model exhibits scalability with large data volumes and supports flexible, secure query types. This framework offers a practical pathway toward HE-enabled secure data retrieval in cloud environments