Task allocation and incentive mechanism design for mobile crowdsensing
University of New Brunswick
Mobile crowdsensing (MCS) is a new paradigm of data collection with large-scale sensing. A group of users with mobile devices (e.g., smartphones, tablet computers, and wearables) are recruited as workers to move around in a specific region and carry out sensing tasks. There are two key problems in MCS, i.e., task allocation problem and incentive mechanism design. In this thesis, we build a MCS framework and provide solutions to its two key problems. Specifically, we focus on (1) the task allocation problem in the static scenarios, in which the information of tasks and workers are known at the beginning of sensing activities; (2) the task allocation problem in the dynamic scenarios when the platform cannot obtain the information of workers before their arrivals; and (3) the incentive mechanism design that motivates workers to participate in the sensing activities. Our proposed MCS framework is associated with two important components to deal with the task allocation problem and incentive mechanism design, respectively. The task allocation problem is considered and formulated as a path planning problem since the tasks in the MCS framework are generally location-dependent. We first plan paths for workers in the static scenarios. Meanwhile, we take two different modes of path planning into account, i.e., the platform-centric mode and worker-centric mode. To solve the path planning problem in these two modes, we propose an evolutionary algorithm and a heuristic algorithm, respectively. Second, we investigate the dynamic task allocation problem. Although the platform has incomplete information of workers, we explore several online algorithms to achieve the satisfactory performance. Third, we design a location-protected and truthful incentive mechanism to motivate workers to move around and accomplish sensing tasks. Based on the results of path planning, we use the auction theory to ensure workers to provide their true private information including costs and task sets. The overall performance of our proposed framework is extensively evaluated through simulations and the simulation results illustrate the effectiveness and efficiency of our solutions in the proposed framework.