Multi-writer Multi-reader Boolean Keyword Searchable Encryption

Abstract

In traditional public key searchable encryption (PKSE), a data owner (writer) utilizes data user’s (reader) public key to build ciphertexts. Thus, to share D data items (W keywords per item) with R readers, a writer suffers from $$O(R \cdot D \cdot W)$$ computational overhead. Researchers then offer numerous schemes supporting multiple readers with optimal overhead, i.e., $$O(D \cdot W)$$ . However, these schemes support either a single-keyword search or multi-keyword search in single-writer multi-reader (SWMR) setting where a writer could share data with the known set of readers. On the other hand, the existing multi-writer multi-reader (MWMR) PKSE offers Boolean search but suffers from $$O(R \cdot D \cdot W)$$ computational overhead for each writer. We observe applications desiring PKSE where writers could share data with the unknown set of readers and readers could perform search across data for Boolean query. Since the existing literature lacks such schemes, we propose a multi-writer multi-reader Boolean keyword searchable encryption (MWMR-BKSE) where the separate sets of registered readers and writers are prepared. Once registered, the writer could compute ciphertexts without knowing the potential readers and the readers could search across data at any time. The computational overhead on each writer is optimal $$O(D \cdot W)$$ . Additionally, a deregistered client would not be able to further write or read data. With MWMR-BKSE, we offer a Boolean keyword search utilizing key policy attribute-based encryption. The security of ciphertext against chosen keyword attack is proved using full security model. With the detailed theoretical analysis and extensive simulations on real-world datasets, we demonstrate the effectiveness of MWMR-BKSE.