MARACrowd: A Multi-Attribute Reverse Auction for Task Allocation in Blockchain-Based Mobile Crowdsensing

Mobile crowdsensing (MCS) is a data-driven application that harnesses the collective intelligence of the crowd for large-scale data collection. Prior MCS systems leverage centralized platforms to interact with users and make task allocation decisions. However, these platforms are susceptible to numerous threats. Although the integration of emergent blockchain technology into MCS may alleviate some of these issues, how to concretely design an optimized, fair, and trusted task allocation scheme remains largely unresolved. This paper proposes a two-stage sealed multi-attribute reverse combinatorial auction framework for task allocation in blockchain-based MCSMARACrowd. Considering the complexity of the task allocation due to the several conflicting decision criteria in the decision making process, the linear weighted sum model is proposed to efficiently allocate tasks. Furthermore, a reliability-aware payment determination model is proposed to incentivize users’ participation and improve data quality. MARACrowd is shown to possess the desirable properties of truthfulness, individual rationality, and computational efficiency. MARACrowd is deployed on an Ethereum test network, and extensive experimental evaluations based on real-world and synthetic datasets demonstrate its feasibility and significant performance.