Duration-Variable Participant Recruitment for Urban Crowdsourcing With Indeterministic Trajectories

Abstract

For urban crowdsourcing applications, the data sensing tasks can be achieved by vehicles traveling on the roads, which can save the expenses on constructing dedicated infrastructures. In this paper, to efficiently handle the crowdsourcing recruitment problem, we propose to recruit participants with the duration-variable principle and prove that it performs better compared to the strategy that recruits vehicles for all required time periods. The duration-variable principle enables recruitment of vehicular sensing resources across different time epochs to maximize the crowdsourcing profits. To ensure the utilization of limited budget, we formulate the duration-variable participant recruitment (DPR) problem with the consideration of indeterministic trajectories to determine which vehicle subset is chosen and how many epochs are provided for each selected vehicle. Since the formulated problem is NP-hard, we propose a two-step DPR algorithm by maximizing the available sensing resource utilization efficiency in each recruitment round, which is shown to be near-optimal and has low computational complexity. Experiments on real traces show that the proposed DPR scheme exceeds three other solutions in providing higher spatial coverage for urban crowdsourcing applications.