Research And Implementation Of User Privacy Protection Algorithm Based On Spatial Crowdsourcing

Crowdsourcing refers to a distributed problem solving mode,which is to find some unknown non-professionals to deal with a complex task through a large public network platform.With the continuous development and improvement of mobile Internet technology and Internet of things technology,some crowdsourcing tasks that rely on users' location information are beginning to emerge,and such crowdsourcing mode is called spatial crowdsourcing.The spatial crowdsourcing platform would acquire the real location of the workers in real time when workers perform a spatial task.And it would cause the sensitive information of workers to be revealed,and thus pose a serious threat to their privacy.Privacy protection and task assignment are two core researches of spatial crowdsourcing.Base on the spatial crowdsourcing model,this thesis analyses the defect of existing correlative algorithm and improve it according to the data characteristic and the service feature of the dispatch system in the real-time vehicle service.The main research contents of this thesis are shown as follows:(1)An algorithm which is called SDP-Grids to protect the drivers' location privacy in real-time vehicle service is proposed,and it realizes differential privacy spatial decomposition.For the lack of dealing with large-scale and skew data in real life,and the disadvantage of not considering the distribution of the original spatial data,this thesis adopt differential privacy Bernoulli sampling to realizes differential privacy spatial decomposition.And it proposes a measure of the sparsity of the spatial data,which make up for the defect of traditional algorithm.Finally,the validity of the algorithm is verified by experiment.(2)According to task assignment of the traditional spatial crowdsourcing and the real-time vehicle service,a driving task allocation scheme on condition of privacy protection is proposed.Then an algorithm about building driving task geocast region is given,which is called BTGR.Traditional task assignment method is always rough and ignores the practical application.This thesis proposes a standard of cell task accept rate by analyzing the influence of the drivers' credit on the completion rate of the driving task.In the end,the effectiveness of the algorithm is verified by experiment.(3)A common model is proposed,which is used to query or publish spatial data.And a model of real-time vehicle service,which is based SDP-Grids,is proposed.And then the application of SDP-Grids and BTGR algorithm in this model is explained.Finally,a vehicle dispatching prototype system is designed based on the above two algorithms.