Collaborative Task Allocation in Mobile Crowd Sensing

Task allocation, i.e., allocating mobile crowd sensing tasks to candidates, attracts more attention in recent years. In the real world, many complicated tasks like troubleshooting, event organization, etc., often require to hire a group of workers with i) group collaboration, i.e., the workers in the group complete the task collaboratively; ii) customized group size, i.e., a certain number of workers according to the task's requirement. However, most of existing task allocation studies neglect the above two aspects. Motivated by this, in this paper we propose a Collaborative Task Allocation (CTA) framework. It includes two stages: firstly, in order to capture collaboration preferences and abilities, it learns the embedding of groups and tasks respectively; secondly, it searches the optimal group for each task that can maximize the overall utility. Extensive experiments are conducted on two real-world datasets. The results show that, compared to the four baselines, the average task execution performance of the proposed method increases up to 155.17%.