Privacy-preserving based task allocation with mobile edge clouds

With the rapid development of mobile crowdsensing applications, task allocation has emerged as a new problem to be solved. Many task allocation strategies have been proposed to select the proper participants to complete tasks. Those methods need participants to submit their real locations to the platform in order to realize the optimal task assignment. However, those traditional task allocation strategies have two weaknesses. First, centralized task allocations result in high computing and communications loads. Second, the exposure of real locations increases participants' concerns regarding location privacy. To address these problems, in this paper, we propose an optimal geo-indistinguishable task allocation (GITA) mechanism using mobile edge clouds. First, the new task that is received by the remote cloud is sent to the mobile edge cloud that is nearest to the task location. Then, the mobile edge clouds serve as distributed controllers to allocate the assigned tasks to the proper candidates. To protect the candidates' real locations, we utilize a geo-indistinguishable mechanism based on differential privacy to preserve location privacy. Specifically, we obfuscate the participants' real locations as disturbed locations, and realize the optimal task allocation based on these disturbed locations. Furthermore, we apply multiobjective mixed integer nonlinear optimization to solve this problem. Finally, extensive experimental results show that, compared with the traditional Laplace mechanism and another privacy-preserving task allocation strategy, the GITA mechanism that is proposed in this paper can decrease users' moving distances and raise the task completion rate. (C) 2019 Elsevier Inc. All rights reserved.