| With the continuous development,innovation and integration of information technologies such as big data,Internet and artificial intelligence,the Internet of Things,Internet of Vehicles and related technologies have been widely popularized and applied,which greatly facilitates people’s lives.At the same time,the access network of various intelligent terminal devices generates a large amount of user data,and various new technologies for in-depth mining and analysis of user data have also emerged.Since the collected user information is often private and sensitive,it is especially important to protect the vast amount of data that users generate on the network every day.In recent years,the development of positioning technology has achieved remarkable results,location-based service systems have been widely used,location data collection is becoming more and more accurate,each intelligent terminal device that can be located more and more frequently reports a large amount of user location data to the server,if this information is used by criminals,it will cause unnecessary losses to the user’s personal and property safety.In today’s network environment,users are increasingly concerned about the privacy of protecting personal data,and only in the local environment can users upload their location data information with confidence.Differential privacy is known for its strict privacy protection conditions,but for protection in the local environment,this article adopts a localized differential privacy approach to protect location data.The biggest difference between localized differential privacy and centralized differential privacy is based on untrusted third-party data collectors,where each user scrubs their own data locally.Aiming at the protection of location data,this thesis studies from one-dimensional location data and high-dimensional location data,and first proposes to use RAPPOR algorithm to protect users’ location data,and make frequency estimation according to the frequency of each location data node.Then,the RAPPOR algorithm is improved,and by measuring the mean squared error,it is proved that the improved algorithm is superior to the original algorithm.Secondly,by integrating the Laplace mechanism into the idea of random response,by estimating the frequency of each location data node,it proves the possibility of this idea,and achieves a relatively good balance between data availability and protecting data privacy.Next,from the perspective of protecting the location nodes on the path,the reality is fully considered,the road network dataset is referenced,and the Laplace noise is added to perturb all the location nodes on the shortest path and protect the path information.Finally,the processing methods of high-dimensional location data are discussed,and the corresponding location protection technology is studied based on the clustering algorithm,and the algorithm is improved under the premise of balancing data availability and privacy protection,and the real data set is introduced,all of which achieve the expected effect. |
