QUALITY-AWARE INCENTIVE MECHANISMS FOR MOBILE CROWDSENSING

The ubiquity of smartphones and their sensors has led to the rise of Mobile CrowdSensing (MCS), the practice of crowdsourcing sensory information from mobile devices. However, little research has focused on the quality of the recruited crowd in MCS, which is the focus of this project. I also consider fine-grained MCS, in which each sensing task is divided into subtasks. I first introduce mathematical models that quantify the quality of a recruited crowd. Based on these models, I present a novel auction formulation for quality-aware and fine-grained MCS that minimizes the service user’s expected expenditure subject to the quality requirement of each subtask. I then discuss how to achieve the optimal expected expenditure and present a practical incentive mechanism, which I show to have the desirable properties of truthfulness, individual rationality and computational efficiency. Finally, I conduct trace-driven simulation using a mobility dataset of San Francisco taxis. Extensive simulation results show the proposed incentive mechanism produces close-to-optimal solutions, noticeably saving expenditure in comparison to two well-designed baseline methods and running time in comparison to the optimal mechanism.