GeoPrivacy Preservation And Data Trust-worthiness In Participatory Sensing

Smart city is the advanced stage of city informatization,which includes smart transportation,smart logistics,smart home,smart health care and application in other fields;Its foundation lies in the "thorough sensing" of all aspects of the city.Sensors based on the Internet of Things obtain massive data from the environment and human activities,which has become the cornerstone of smart city applications.However,it costs hugely to deploy large-scale IOT equipments and software in urban environment.In view of this,people can perceive the data of the physical environment,social environment and individual entities state of cities,through a combination of different intelligent mobile terminals with corresponding applications in the wake of participatory perception Such low-cost data collection method of strong expansibility reflects the wisdom of smart city construction.At present,participatory sensing has become an effective paradigm for large-scale data collection,and it has been faced with such major challenges as task allocation and participant selection,incentive mechanism design,data security,data redundancy and quality evaluation,and malicious behavior detection.Data security is related to "quality" of the participatory sensing system.As two important components of data security,geoprivacy and data trustworthiness have a significant impact on effective data collection and publishing.In the absence of effective protection mechanism,sharing sensing data including participants' spatial and temporal information may lead to the disclosure of their privacy,which in turn will discourage participants from the perception task.Meanwhile,due to the openness of participatory sensing system,malicious participants may forge or tamper data for rewards,undermining data quality and analysis.Therefore,based on the research needs of smart city participatory sensing system and the basic concept of geographic information,this dissertation redefines geoprivacy and data trustworthiness,and carries out corresponding research on geoprivacy protection and data trustworthiness in participatory sensing.The main contributions are as follows:1.Pathswap,a geoprivacy protection method based on the confusion of sub trajectories in participatory sensing,is proposed.This method combines the mix-zones and k-anonymity model.A communication protocol between an anonymous server and an application server is designed,without introducing a trusted third party.By hiding the corresponding relationship between identity and trajectory,identities are exchanged in the anonymous server.As such,the bonds between the participant's identity and its original trajectory is cut off.The results of simulation tests on three real trajectory data sets show that Pathswap can better resist the re-identification attack based on behavior pattern,achieve the purpose of protecting the geoprivacy of participants,and retain the characteristics of dataset aggregation statistics to the greatest extent,compared with the existing privacy protection methods based on differential privacy and k-anonymity model,2.A quantifying framework for trajectory privacy protection based on unified measurement is proposed to solve the evaluation of different privacy protection methods.In the framework,a unified privacy metric is defined,which integrates the attacker's background knowledge and attack models.This privacy metric replaces the specific metrics of different privacy protection methods with an integration of trajectory similarity and user uniqueness,and achieves a unified evaluation of privacy protection levels.Relevant experiments show that the framework and the corresponding unified privacy metrics can effectively assist researchers in designing or selecting appropriate trajectory privacy protection methods in accordance with the requirements of protection and data availability.3.TPSense,a vehicle participatory sensing framework based on fog architecture is designed to achieve the dual goals of traffic event-report trustworthiness evaluation and participant geoprivacy protection.Different from the traditional data trustworthiness evaluation method based on the reputation value of nodes,TE-EM algorithm in the framework can be employed to transform the trustworthiness evaluation problem of traffic event-report into the maximum likelihood estimation problem,and solve the problem through expectation maximization algorithm.Furthermore,anonymous signature technology is used to generate a pseudonym to replace the vehicle's real identity so as to ensure that the vehicle node is not associated and tracked,realizing the geoprivacy protection of the vehicle node.Simulation tests have verified the effectiveness of the scheme.4.Trust E-VPS,an integrated data trustworthiness evaluation framework based on trust model is designed.It integrates common attack models,different types of trust models and metrics.The efficiency and performance of various evaluation methods are evaluated by constructing perception scenarios and attacks in different context.The implementation of Trust E-VPS is based on SUMO,an open-source traffic simulation software and on OMNe T++,an open-source wireless communication simulation software which presents strong applicability and scalability,and provides smart cities planners and designers with a perfect trust model and a design and test environment for trustworthiness evaluation method based on the trust model data.5.PUEP,a trajectory privacy measurement system based on workflow.PUEP system defines trajectory data preprocessing,privacy protection method,privacy and data utility metrics as operators that can be repeatably employed.Then,the experiment task is described in the form of directed acyclic graph,and finally the automatic execution of the experiment task is realized.Through PUEP,researchers can define and deploy experiments directly,and realize the routing management of design,evaluation,experiment results collection and analysis of privacy protection methods.PUEP system also allows users to develop new trajectory privacy protection methods and metrics.