Research On Privacy Protection In Mobile Crowd Sensing Networks

Mobile Crowd Sensing(MCS)networks,as an important means to achieve a comprehensive perception of the Internet of Things,use the mobile terminal devices carried by lots of ordinary users as a basic sensing unit,and adopt var-ious communication manners to collaborate,in order to conduct sensing task distribution and sensing data collection and analysis,and finally complete the large-scale and complicated social sensing tasks.The MCS networks charac-terize low cost,strong dynamic and good scalability,making themselves much easier to cover the entire physical world and collect a wide range of data.But the MCS networks also bring new challenges.The basic premise of MCS ap-plications is that people would contribute their collected sensing data,however these sensing data may carry lots of individual sensitive information,making users risk personal privacy disclosure.Therefore,it is a very important problem to protect users' privacy in MCS applications.In this thesis,we focus on the two aspects:location privacy and data pri-vacy in the sensing data-collecting and information service-providing phases of MCS applications.Our main contributions are as follows:(1)Participant density-independent location privacy protection.To protec-t location privacy in the sensing data-collecting phase,we propose a location privacy-preserving scheme based on the combination of multiple anonymities and multi-party authentication.In order to avoid the issue that the upload-s from areas with low density participants might leak a participant's location privacy,we present a mechanism of reporting participants' sensing data by leveraging multiple pseudonyms in mobile terminals.To further handle the Sybil attacks from malicious participants by using the multi-pseudonym re-porting mechanism,thus polluting the aggregate statistical results,we design a multiparty cooperation-based sensing data verification method on the service platform,which can effectively detect participants' sensing data with differ-ent pseudonyms.The proposed scheme can protect the location information of the contributors from revealing sensitive information,and also can obtain their more accurate location information,making the aggregate statistics more accurate and valuable.(2)Data privacy protection in mobile sensing environment.Aiming at the problem of sensing data privacy protection in mobile sensing environment,we propse a data privacy-preserving scheme based on dynamic group manage-ment.At first,considering the dynamic changes of participants and the limited energy of mobile terminals in the mobile sensing environment,we design a ring-based grouping lightweight encryption technology to encrypt the upload-ed sending data.Then,taking into account the insecure open environment,we propose a data aggregation integrity verification protocol to check the correct-ness of statistical results.In addition,due to the low QoS in wireless networks,we leverage the future message buffering mechanism to guarantee fault toler-ance of data aggregate statistics and integrity verification.(3)Differential privacy protection on correlated sensing data.We first investigate the influence of sensing data correlation on differential privacy pro-tection mechanism in the context of MCS networks.Then we study the problem of the privacy protection of correlated sensing data from two different perspec-tives,and propose two differentially private data perturbation methods based on data correlation model.From the viewpoint of a protector,we model the data correlation between tuples as a Bayesian network,and then start from the definition of differential privacy to deduce the scale parameter ? based on the Laplace perturbation mechanism,thereby utilizing it to propose a new pertur-bation mechanism to satisfy differential privacy.From the viewpoint of an attacker,we exploit Bayesian differential privacy as privacy protection model.Based on the Gaussian correlation model to describe the data correlation,we explore the influence of the maximum correlated group on the Bayesian differ-ential privacy leakage,and then give a corresponding perturbation algorithm.(4)Design and implementation of privacy protection for air quality mon-itoring system.In order to verify the feasibility of the proposed schemes,we design and implement the privacy protection module for the air quality mon-itoring prototype system.After users obtain the PM2.5 of their location,the system provides data or location privacy-protection strategies according to the proposed schemes,so that the users can securely report their sensing data to the service platform for aggregate statistics.The experimental results show that the statistical results under privacy-preserving strategies achieve small error com-pared to the direct ones,which can meet the needs of practical applications.In summary,this thesis proposes a series of models and methods in terms of location and data privacy protection.Moreover,we demonstrate their ef-fectiveness and feasibility via theoretical analysis,extensive simulations and prototype system.All of these can provide significant theoretical and technical support for the wide application of the MCS networks.