PDMA: Efficient and privacy-preserving dynamic task assignment with multi-attribute search in crowdsourcing

Crowdsourcing leverages distributed mobile devices for task allocation, significantly reducing service costs. However, existing schemes face three major challenges, i.e., data privacy leakage, focusing just on single-attribute tasks, and the inability to accommodate dynamic task updates. To address these issues, we propose a privacy-preserving dynamic multi-attribute task assignment scheme (PDMA). PDMA supports multi-attribute range searches by incorporating spatial, temporal, and keyword constraints. It introduces a hilbert attribute tree (HRAT) for efficient query of multi-attribute tasks and utilizes hilbert R-trees and counting bloom filters (CBF) to facilitate dynamic task updates. To preserve the privacy of spatial and temporal attributes, PDMA integrates the improved symmetric homomorphic encryption (iSHE) scheme, while hash functions preserve the CBF for keyword privacy. Additionally, we propose a secure ternary match protocol (CTP) and a secure subset query scheme (Ssub), which combine iSHE-based ciphertext comparison protocols with simulated ternary content addressable memory (TCAM) to accelerate keyword subset matching. Security and performance analysis demonstrate that PDMA achieves the chosen-query attack security (CQA2-security) and is both practical and efficient.