Differentially private spatial crowdsourcing

Abstract

In recent years, the popularity of mobile devices has transformed spatial crowdsourcing into a novel mode for performing complicated projects. Workers can perform tasks at specified locations in return for rewards offered by employers. Existing methods ensure the efficiency of their systems by submitting the workers’ exact locations to a centralized server for task assignment, which can lead to privacy violations. Thus, implementing crowsourcing applications while preserving the privacy of workers’ location is a key issue that needs to be tackled. During the process of task assigning and task reporting, workers and requesters are usually required to reveal their locations to potentially untrustworthy entities such as the SC-server, other workers and other requesters, or the server may collect and release the location data of workers and requesters for further analysis, leading to possible privacy breaches. In recent years there have been a number of proposals to provide the privacy preserving capability for SC applications, such as allowing the release of spatial datasets while preserving privacy. This chapter first surveys the current attempts to solve the location privacy problem in SC, and then presents a novel method for reward-based SC with a differential privacy guarantee. A reward allocation mechanism is proposed to adjust each piece of the reward for a task using the distribution of the workers’ locations. Through experimental results, it shows that an optimized-reward method is efficient for spatial crowdsourcing applications.