An Anonymization Method Based On Anatomy And Reconstruction In LBS Privacy Preservation

With the rapid development of mobile Internet in recent years, location-based service (LBS) has been widely used. LBS brings great convenience in people’s daily live, however it also lead to privacy problems. Therefore, privacy problems in LBS has gained more and more attention from the academia and industry. Anonymity is an effective way to preserve privacy. The main idea of anonymity is transforming the user’s LBS query requests to make sure that attackers cannot identify the user’s location. This thesis investigates the LBS privacy issues from the views of anonymized technologies, anonymity models, anonymized algorithms and quality assessment models. The main contributions are as follows:(1) We introduce the anatomy technology into the field of LBS privacy preservation. We propose a novel anatomy and reconstruction technology for LBS privacy preservation.The existing anonymized methods usually adopt spatial-temporal cloaking technology, which is inefficient. Response with long time delay will result in low quality of services. Therefore, based on the research of anatomy technology which has been used in micro data privacy preservation, we propose an anatomy and reconstruction technology. The technology firstly partitions LBS queries into some equivalence classes satisfying anonymous constraints, then each LBS query in equivalence class is decomposed and new anonymous query sets are reconstructed.(2) We propose a series of anonymity models to satisfy varieties of privacy requirements. We also propose some anonymized algorithms to implement these anonymity models. In addition, some assessment models oriented to anatomy and reconstruction technology are proposed from the view of security and QoS (quality of service). The simulation results illustrate the effectiveness of the anatomy and reconstruction technology.(3) To resist exclusion attacks, we propose (k, l, α,β)-anonymity model. The model not only can satisfy the query diversity and location diversity, but also can control the ratio of invalid queries. We also propose anoymized algorithm to implement the (k, l, α,β)-anonymity model. Finally, we analyze the (k, l, α, β)-anonymity model and its anonymized algorithm from the view of security and QoS.