Abstract
In mobile crowdsourcing, workers are financially motivated to perform as many self-selected tasks as possible to maximize their revenue. Unfortunately, the existing task scheduling approaches in mobile crowdsourcing fail to consider task execution duration and do not scale for massive tasks and large geographic areas. In this article, we propose a novel framework, Turbo-GTS, in support of large-scale geo-task scheduling, with the objective of identifying an optimal task assignment for each worker to maximize the total number of tasks that can be completed for an entire worker group, given the geographic locations of each task and each worker. Since the exact solution to the geo-task scheduling problem is computationally intractable, we first propose two sub-optimal approaches (least cost neighbor with particle filtering and non-urgency degree particle filtering with iterative clustering) based on particle filtering and DBSCAN for the single-worker geo-task scheduling problem. We then extend our work to solve the multi-worker geo-task scheduling problem by proposing two space partitioning-based methods (QT-NNH and QT-NUD), which leverage point-region quadtree to ensure workload balancing. The effectiveness and efficiency of the four proposed approximate solutions are verified by our extensive experiments using both real and synthetic data. Compared to state-of-the-art approaches, our proposed solutions are able to return a higher number of completed tasks for the worker group while reducing the computation cost by up to three orders of magnitude when coping with massive tasks distributed in large geographic areas.