Exploiting Data Reuse in Mobile Crowdsensing

Mobile crowdsensing emerges as a promising sensing paradigm through leveraging the diverse embedded sensors in massive mobile devices. A key objective in mobile crowdsensing is to efficiently schedule mobile device users to perform multiple sensing tasks. Prior work mainly focused on the interactions between the task layer and the user layer, without considering the similarity of tasks' data requirements and the heterogeneity of users'sensing capabilities. In this work, we propose a three-layer data-centric crowdsensing model by introducing a new data layer between tasks and users, which allows us to effectively leverage both the task similarity and the user heterogeneity. We formulate a joint task selection and user scheduling problem on top of the data layer, aiming at maximizing the social welfare. This problem is difficult to solve due to the combinatorial nature as well as the two-sided private information of tasks and users. To address both issues, we propose a two- sided randomized auction mechanism, which is computationally efficient, individually rational, and incentive compatible in expectation. Simulations show that (i) the proposed randomized auction can achieve 90% of the maximum social welfare (benchmark), and (ii) the social welfare gain due to data reuse increases with the task similarity and reaches up to 1300% in our simulations.