Efficient Online Vehicle Recruitment Based on Deterministic Trajectory in Mobile Crowd Sensing

Vehicle-based mobile crowdsensing (MCS) is a novel paradigm for data collection that recruits vehicles over a large geographic area to fulfill complex sensing requirements. In reality, vehicles are dynamically involved in recruitment over time, which requires an online manner to assess and recruit vehicles. Compared with offline recruitment, online recruitment is executed with the context that only includes trajectory information of vehicles connecting to the MCS at the current time, which makes the recruitment decisions in real-time. In this paper, the online vehicle recruitment mechanism based on deterministic trajectory is proposed. This mechanism periodically analyzes the spatio-temporal availability of trajectory to address the problem of matching vehicles with their assigned tasks while optimizing recruitment costs. Specifically, we study the interaction between trajectory and task, and establish the assessment to analyze the spatio-temporal availability of trajectory in terms of coverage and redundancy. Finally, we design and implement an efficient algorithm (VRP-PC) based on the online vehicle recruitment mechanism and compare its performance with an ideal recruitment algorithm (VRP-FO) in simulations. The performance of the VRP-PC algorithm is close to the ideal algorithm in terms of coverage, while recruiting fewer vehicles.