Research On Data Protection Of Crowdsensing Based On Differential Privacy

Mobile Crowdsensing network is a kind of sensing network which is composed of many mobile devices that integrate a large number of sensors by utilizing a large number of ordinary users.As basic sensing units,ordinary users' mobile devices perform conscious or unconscious cooperation through the mobile Internet to complete task assignment and collection of sensing data.However,when participating users use their actual location to report sensing data,their location privacy information is inadvertently compromised.In order to solve this problem,this thesis introduces ?-differential privacy,which provides theoretical guarantee for the participant's location privacy.Firstly,this thesis introduces a brief overview of the research background,research status and research significance of crowdsensing and location privacy protection respectively,and briefly introduces the definition of differential privacy,noise addition mechanism and the definition of sensitivity.Then,this thesis introduces the composition and incentive mechanism of the crowdsensing network,as well as the basic methods of location privacy protection,which makes this thesis have a sufficient theoretical knowledge reserve.Based on the above research,this thesis proposes a location differential privacy protection scheme for crowdsensing(LDPP),which has strong privacy protection and excellent data availability at the same time.The main steps of the scheme are to divide the target area into multiple areas of the same size,model the user position as a matrix,calculate the optimal obfuscated matrix using historical data based on differential privacy,and map the real position uploaded by the user to another dummy position.And a multi-dummy setting method based on k-anonymity is used to invest multiple dummy points in the area and anonymize the user before uploading the data.In addition,it also defines a distance-based location privacy protection strength measurement algorithm to quantitatively compare the privacy protection strength of each scheme,and compare the data quality loss of each scheme by calculating the average absolute error.In order to increase the availability of location differential privacy protection scheme in the crowdsensing network,improve its practicability and scalability,and realize the interactivity of data processing and publishing,this thesis designs a prototype system for location privacy protection in the crowdsensing.In addition,it evaluates the location privacy protection scheme,conducts a series of simulation experiments by adjusting the parameters(number of participants,privacy budget andnumber of divisions),and compares this scheme with the others.Both privacy protection strength and data availability prove the superiority of this scheme.Finally,this thesis summarizes the scheme,proposes the defect and looks forward to the future work.