User Privacy Protection For Distributed Storage In Location-based Service

With the popularity of cellular networks,GPS and other technologies,people can use mobile phones at any time and any place and request services from the location based-service(LBS)provider,which greatly promotes people's lives.Many scheme have been proposed,but they still don't do a good job of balancing the intensity and efficiency of privacy protection.It remains a very practical research topic that how to prevent service providers or other attackers from accessing sensitive user information without affecting the users' use of fast and convenient services.This subject uses encryption to protect user privacy,at the same time uses distributed storage to reduce system overhead.In addition,the homomorphic encryption and tree structure are used to solve the difficult problem of k-nearest neighbor query.In chapter 3,based on distributed storage and the quadratic residual theorem,a privacy protection scheme for location-based users is proposed with both security and effectiveness.Although the traditional private information retrieval(PIR)encryption method brings high communication cost and computing pressure to the LBS server side,it has the advantage of no need of trusted third party and high privacy protection intensity.What's more? the distributed storage system architecture can greatly reduce the network overhead and reduce the risk of the LBS server identifying the user's actual location,because the cache proxies can respond to the request in advance when the user's request does not arrive at the LBS server.In order to allow the intermediate cache proxies to correctly respond to encrypted requests without knowing the user's request,we constructed an index table to solve this problem and designed a scheme for distributed storage of location-based service user privacy protection.Simulation results show that the scheme can effectively reduce user communication and LBS server computing overhead.In chapter 4,the difficult problem of k-nearest neighbor query based on cryptography is studied,and the k-nearest neighbor query algorithm DPIR-NN is proposed,which is suitable for the designed scheme.Instead of using the traditional Hilbert curve to store the information of the points of interests in the target map,DPIR-NN uses the Z-order curve and quad-tree to store these location information.In this way,it not only supports distributed storage but also improves the efficiency of localization.At the same time,the algorithm uses the homomorphic encryption algorithm ASM-PH that supports add and multiplication homomorphism to keep the query secret,which can save the user's cost by calculating the encrypted content with the LBS server or the storage devices when only the user knows the encryptedcontent.Experimental results show that the algorithm has high query accuracy and low user encryption and decryption cost.