Research On Crowdsourcing Privacy Protection Method Satisfying Local Differential Privacy

Space crowdsourcing means that task performers go to specific locations to complete crowdsourcing tasks.This working model has been widely studied and used.The space crowdsourcing work model includes two links: the first is the worker recruitment link,that is,the worker user submits personal location information to the server for registration,and there will be specific location privacy leaks;the second is the task distribution link,that is,the space crowdsourcing server distributes tasks through tasks When the algorithm distributes tasks to worker users,there will be problems such as privacy leakage of the worker's location and excessive time overhead.Based on the problems of the above-mentioned crowdsourcing work model,we conducted research and proposed solutions respectively.The specific work is as follows:(1)Research on the problem that the traditional differential privacy mechanism for location data protection is too sensitive,leading to insufficient data availability.It is proposed to apply the geo-indistinguishability to crowdsourced location data protection,introducing the concept of distance and users Privacy level factor.By setting the privacy level of the user,the scope of the generated virtual location can be restricted,and the privacy of the user's location can be personally protected.According to experimental verification,the data has higher usability without revealing the real location of the user.(2)In the process of spatial crowdsourcing,the problem of privacy leakage of user's regional position during regional segmentation is studied,and an adaptive two-level work area grid partition method is proposed,and a two-level grid partition granularity function is constructed,which calculates a reasonable two-level grid partition granularity according to the difference of regional population density,so as to achieve personalized partition.At the same time,Laplace mechanism is used to protect the workers' position count in the region.This paper studies the waste of privacy budget based on Laplace mechanism,and proposes an adaptive proportional distribution method of privacy budget.According to the experimental results,according to the different situation of regional workers' counting proportion,different privacy budgets are allocated to each region,thus achieving personalized privacy protection effect.(3)Studying the problem that the spatial crowdsourcing server spends too much time processing large amounts of data,this paper proposes to apply greedy algorithm to the work area,and adopt the optimal stopping theory to match tasks by introducing the factors of task demand credibility and task distance.According to the experimental verification,it can reduce the task allocation time of the task server while ensuring its task allocation efficiency.