Research On Data Privacy Protection Method For Vehicle Swarm Sensing

The unprecedented popularity of mobile smart devices has promoted a promising computing paradigm,namely Mobile Crowdsensing(MCS),where people can share personal data about their surroundings with others.As a fast,simple,and cost-effective method for solving large-scale social problems,MCS is widely used in many fields such as environmental monitoring,map construction,and public safety.As one of the latest and most promising solutions in MCS,smart vehicles are designed to remotely collect sensing data(such as time stamps,GPS locations,speeds,etc.)from vehicles participating in sensing tasks.In this process,it is only necessary to explore the sensing technology embedded in the smart vehicle to deploy a powerful monitoring system without extensive installation of other infrastructure.With its wireless communication function,vehicle participants can directly provide the perception data they collect to the cloud service provider for query posting of perception tasks.However,while the Vehicular Crowdsensing brings convenience to people's daily travel,its perception data contains a large amount of private information related to its identity.If these data are directly released without processing,it will pose a serious threat for participants.Therefore,it is extremely urgent to protect the privacy of participants in Vehicular Crowdsensing.In the context of Vehicular Crowdsensing,this thesis combines the theory of differential privacy to study how to effectively protect the privacy of vehicle participants in different situations.The main work is as follows:(1)Considering that when a cloud service provider processes a large amount of sensed data,in order to divide the sensed data attributes into a cluster,the statistical analysis of the sensed data of each cluster is used to query and publish,designed a vehicle sensory data clustering algorithm that satisfies differential privacy.In the algorithm,the Laplace mechanism of differential privacy is combined,and the DBScan algorithm based on density is improved.Noise is injected into the Euclidean distance of every two vehicle data objects,and the sensory data attributes are similarly divided into a cluster.Finally,experiments prove that the algorithm can provide participants with privacy protection and improve the effectiveness of the clustering algorithm.(2)Considering the high latency and low efficiency caused by uploading a large amount of sensory data directly to the cloud service provider.The fog node is introduced,and a fog-based privacy protection vehicle sensory data aggregation algorithm is designed.After the participants of the vehicle collect the sensory data,they will deal with their sensory data to a local differential privacy disturbance,and then upload the disturbed data to the fog nodes in the area.The fog nodes will analyze and process the disturbed data in the area.Then,upload the aggregated results to the cloud service provider.Finally,the design of the experiment shows that the algorithm can provide privacy protection while reducing the communication overhead between the cloud service provider and the vehicle,and make the query response more efficient and faster.