Local Differential Privacy Location Protection Method Based On Hilbert Encoding

With the rapid development of positioning technology,various location-based big data services are constantly emerging,bringing convenience to people’s lives while also posing a significant risk of privacy leakage.In traditional location privacy protection methods,users first upload their precise location to a third-party data collection platform,which processes privacy and submits it to the location service provider.However,in practical applications,third-party data collection platforms still face the risk of privacy leakage.Only by ensuring that user privacy is not compromised can users be willing to share location information.Therefore,local differential privacy technology has emerged,which allows location data to be perturbed before leaving the user device,avoiding the disclosure of real data.This thesis designs a new method for protecting user location privacy based on local differential privacy technology.The main work is as follows:1.To overcome the shortcomings of complex encoding mechanisms and the low availability of location data in existing two-dimensional local differential privacy protection methods,a new method based on Hilbert encoding is proposed.This method first uses a uniform grid structure on the server to divide the entire two-dimensional plane space and uses Hilbert encoding to encode the grid position where the user is located,achieving dimensionality reduction of the location data.Then,the user performs random response perturbation processing on the Hilbert encoding of their own grid based on the local differential privacy model,achieving privacy protection at the original location.Finally,the server collects a large number of user disturbance location codes and performs Hilbert decoding to determine the grid location of the users,achieving statistical analysis of the number and distribution density of users.Experiments and analysis of real location data sets have proven that this method outperforms other existing methods regarding location data availability and operational efficiency when the differential privacy parameters are the same.2.To address the problem that existing local differential privacy protection methods are challenging to apply to three-dimensional spatial scenes directly and cannot meet users’ personalized privacy requirements,a three-dimensional spatial personalized local differential privacy protection method based on Hilbert coding is proposed.The method first performs three-dimensional uniform segmentation by the server according to the spatial coverage,then performs three-dimensional Hilbert coding on the cell blocks obtained from the spatial segmentation and sends them to all users.When users use the location-based service,they first determine the spatial cell block they are in based on the real location and the parameters sent by the server;then,they find the corresponding three-dimensional Hilbert encoding and use the personalized perturb mechanism proposed in this thesis to perturb it;finally,they submit the perturbed encoding to the server instead of the original location.The server receives the perturbed code from the user and decodes it using the three-dimensional Hilbert decoding algorithm,then determines the spatial unit block in which the user is located by combining it with the initial spatial segmentation structure and providing various location-based services to the user accordingly.Experiments with different spatial location data sets prove that the method can reduce the quality loss of spatial location services,improve the operational efficiency of the spatial location perturbation algorithm,and maintain good availability of spatially perturbed locations.