Research On Location Privacy Protection Based On Semantic Diversity

With the rapid development of space positioning technology and mobile communication technology, Location-Based Services(LBS) has gradually entered our daily life. Mobile users can use different LBS services anytime and anywhere nowdays. LBS has brought great convenience to people’s daily life, but users are also facing the risk of leakage of location privacy when they enjoy the convenience of LBS. This has aroused the concern of many scholars, and has become one of the hot topics in the research field of privacy protection.In this thesis, we analyze and summarize the existing location privacy protection methods, and points out the advantages and disadvantages of these methods. We anylyze the demand of LBS system, and we designed the privacy protection system framework. In the past, few location privacy protection methods consider location semantic attacks of the mobile user, in this thesis we introduce the concept of semantic location(SL), use SL’s access time distribution for SL’s semantic classification, and the semantic diversity location privacy protection mechanism is proposed in this thesis. We force anonymous server need to satifis L semanic diversity and K-anonymity when generate anonymous area for a mobile user. In order to accelerate the computation of this algorithm, we use the Voronoi Graph partition and RTree Index, the traditional K-anonymity algorithm is extended to support the region K anonymous.We implement this semantic diversity algorithm on a on a recommendation LBS Application, and several location privacy protection methods are deployed on this APP. We made this APP to be a compare platform of various anonymous location privacy protection methods. At last, we use the classical software Network Generator to generate experiments dataset for measurement of semantic diversity of location privacy protection method we proposed. The experimental results of semantic diversification of location privacy protection algorithm we proposed shows it has good performance on both computing time and anonymous region generation.