Hardening Distributed and Encrypted Keyword Search via Blockchain

Distributed storage platforms draw much attention due to their high reliability and scalability for handling a massive amount of data. To protect user and data privacy, encryption is considered as a necessary feature for production systems like Storj. But it prohibits the nodes from performing content search. To preserve the functionality, we observe that a protocol of integration with searchable encryption and keyword search via distributed hash table allows the nodes in a network to search over encrypted and distributed data. However, this protocol does not address a practical threat in a fully distributed scenario. Malicious nodes would sabotage search results, and easily infiltrate the system as the network grows. Using primitives such as MAC and verifiable data structure may empower the users to verify the search result, but the robustness of the overall system can hardly be ensured. In this paper, we address this issue by proposing a protocol that is seamlessly incorporated to encrypted search in distributed network to attest and monitor nodes. From the moment a node joins the system, it will be attested and continuously monitored through verifiable search queries. The result of each attestation is determined via a standard quorum-based voting protocol, and then recorded on the blockchain as a consensus view of trusted nodes. Based on the proposed protocols, malicious nodes can be detected and removed by a majority of nodes in a self-determining manner. To demonstrate the security and efficiency, we conduct robustness analysis against several potential attacks, and perform performance and overhead evaluation on the proposed protocol.