Research On Privacy Protection Methods In Mobile Crowd-sensing Environment

With the continuous maturity of sensor technology,the development of Internet technology and the widespread popularity of smart phones,Mobile Crowdsensing(MCS)has developed rapidly in recent years.Mobile Crowdsensing refers to a participatory and interactive perception network formed by the existing mobile intelligence devices of the participants around the world by combining the idea of crowdsourcing with the perception ability of the mobile intelligent devices themselves.And the perception task is released to the participants in the perception network to complete the task perception,thereby helping the public,professionals,government agencies,etc.to aggregate data,statistical analysis information and share knowledge.Mobile Crowdsensing not only provides convenience for people's daily life,but also makes people travel faster.However,new privacy and security challenges ensue.Mobile Crowdsensing applications require a large number of participants to participate in the collection of sensory data,and these sensory data will carry the participants' personal sensitive information and expose them to the risk of personal privacy leakage.As people pay more and more attention to personal privacy issues,the development prospects of Mobile Crowdsensing applications are worrying.Therefore,the key issue that restricts the development of Mobile Crowdsensing applications is the privacy protection of participants.Based on the theory of differential privacy,this thesis makes an in-depth study of privacy protection technologies in mobile swarm intelligence perception from three aspects:perception data aggregation,personalized participant selection,and trajectory release.The main contents of this thesis are as follows:(1)Considering the risk of privacy leakage of participants' personal information and perceived data,combined with differential privacy technology to improve the privacy protection method in MCS.Aiming at the problem of perceptual data aggregation,research the aggregation method based on noise perturbation and noise filtering.Aiming at the issue of participant selection and privacy protection,research on personalized participant selection and trajectory privacy protection method based on incentive mechanism,trajectory compression and noise disturbance.On the basis of satisfying differential privacy,the above method is guaranteed to have good data utility.(2)Aiming at the problem of privacy leakage of participant positions and perceived data on untrusted MCS platforms,a Mobile Crowdsensing perception data aggregation algorithm DP-DAWL based on location obfuscation was proposed.The algorithm optimizes the K-means clustering algorithm by adaptively selecting the optimal group division number by the contour coefficients,and uses the group center to obfuscate the participants' real location data.Then,based on the different population density of participants,different perturbation noises are added to the perceptual data to achieve privacy protection.Finally,the system noise of the Kalman filter algorithm is optimized on the MCS platform to aggregate and analyze perturbed perceptual data.The experimental results on the six synthetic datasets show that DP IPTM has good data utility on the basis of achieving privacy protection.(3)Aiming at the privacy disclosure problem of trajectory data publishing in Mobile Crowdsensing,a personalized participant selection incentive mechanism that maximizes the perception area and its trajectory privacy protection method DP_IPTM was proposed.This algorithm first proposed the personalized participant selection incentive mechanism MPMA with the maximization of perception area.This mechanism selects the candidate participant set based on similarity and uses the greedy and knapsack fusion algorithm to select the optimal participant set based on the maximization of perception area.Then combined with Douglas-Puke algorithm and differential privacy technology,an offline compression algorithm for participant trajectories that meets differential privacy in background execution scenarios is proposed;Finally,by adaptively calculating the threshold value to optimize the sliding window algorithm and combining the Laplace noise mechanism,an online compression algorithm for participant trajectories that meets differential privacy in mobile execution scenarios is proposed.Experimental results on the synthetic datasets and real dataset GeoLife GPS Trajectories show that DP_IPTM has good data utility on the basis of achieving privacy protection.