Adaptive Assignment for Quality-Aware Mobile Sensing Network with Strategic Users

Mobile crowdsourcing with smartphones is a novel paradigm that collects distributed data efficiently. Millions ofcrowdsourcing tasks should be assigned and performed every day. In order to give good service quality for mobile sensing applications, a practical assignment approach is indispensable for a task platform to select appropriate users. This paper considers heterogeneous task assignment problem in the mobile sensing network that cannot label users. We consider the following scenario: a requester posts a set of tasks onto the platform, the platform selects a subset of proper smartphones to complete the different kinds of tasks under a budget constraint, all users inthese transactions are strategic for maximizing their individual utility. The assignment approach should be designed to fight against strategic behaviors of smartphone users. We address this problem by introducing a novel mechanism for mobile sensing assignment based on the multi-armed bandit problem. Moreover, we present an adaptive algorithm to assign users according to the properties of task and which task they are good at. Finally, we evaluate the performance of our mechanism based on a real dataset, and the results demonstrate that proposed algorithm achieves better in high-quality assignment.