Personalized Privacy-Preserving Task Assignment In Spatial Crowdsourcing

Through spatiotemporal crowdsourcing,task requesters submit crowdsourcing tasks which are related to location and time to crowdsourcing platform,crowdsourcing server assigns tasks to a group of workers,and they will actually go to the task location to perform tasks.However,current task allocation schemes require worker and requesters to disclose their locations and / or tasks to untrusted crowdsourcing servers to assign tasks to workers.But this kind of information disclosure will more or less lead to the disclosure of users' privacy.People's concerns about privacy leakage may lead to their unwilling to participate in crowdsourcing.Based on the users' demand for privacy protection,this paper proposes a three-stage assignment model,which assigns spatiotemporal tasks to the workers online while protecting the location privacy of workers and spatiotemporal crowdsourcing tasks.According to the indistinguishability of geographical location,the location of tasks and workers is disturbed.Due to the poor performance of using the disturbed location directly,this paper quantifies the reachable probability between the task and the worker for the disturbed location of the worker and task.This paper quantifies the empirical model of the reachability of worker tasks,and proposes a task allocation algorithm based on the reachability rate,which can achieve a balance between the number of tasks completed,worker movement distance and system overhead.However,different data owners have different expectations for privacy protection.Users can obtain other benefits by reducing their needs for privacy protection,such as distribution success rate.Therefore,this paper allows worker and tasks to have different privacy protection levels.Worker and task requester can adjust their privacy protection policies to meet the balance between the success rate of task allocation and privacy protection needs.The allocation algorithm in this paper is still competent in the context of personalized privacy.Extensive experiments on real data sets show that the proposed three-stage assignment model and the assignment algorithm based on the reachability rate can disclose the least task location and the least worker location without significantly sacrificing the total number of assigned tasks under the scenario of personalized privacy protection or non personalized privacy protection.