Lightweight verifiable blockchain top-[formula omitted] queries

Abstract

Blockchain has been exploited in many applications as a fundamental technology to construct trust and share data among multiple participants. A user with limited resources who runs a light node fetches data records stored on the blockchain by requesting a full node that maintains the complete blockchain data. As a type of broadly used query, top-k queries which ask for data records with the highest k values in a certain region can be useful in many blockchain scenarios. However, untrusted full nodes may return incorrect top-k results. In this paper, we take the first step toward addressing the problem of verifiable blockchain top-k queries without the help of a trusted third-party. First, we design a novel Sorted Merkle tree-based authenticated data structure and based on it we propose a new index named inverted SMT index that supports verifiable top-k query with specific keyword conditions. Second, we propose a fundamental strawman design to support verifiable blockchain top-k queries with arbitrary query regions. We further introduce two improved schemes, named segmentation design and sliding-window design, that overcome the efficiency obstacle of the strawman design. Finally, we confirm the efficiency of our solutions through an extensive experimental evaluation.