An Adaptive Learning Model for k-Anonymity Location Privacy Protection

Abstract

Location based services (LBS) and the recent awareness towards their privacy threats have kindled the research in providing state of the art approaches and techniques to preserve the user location privacy. Most of these approaches make use of the k-anonymity model to provide personalized location privacy. Through personalization, a k-anonymity model is able to achieve privacy based on the input user profile and can even accommodate changes to user's privacy preferences at per-query granularity. Though this is progressive towards providing user with more control over their location privacy, even the most privacy-centric users might overlook some privacy issues due to complexity in tracking their privacy preferences at a per-query basis. The main goal of this research is to develop a framework that would help users to choose and manage their privacy preferences effectively and to obtain context-based privacy from the anonymizers. Based on analyzing a set of factors that generally influence the choice of privacy profile, a learning model is constructed to help users to make right decisions in protecting their location-based privacy. As the learning model evolves, it will manage different privacy preferences of users for different contexts with minimum user intervention and therefore prevent them from privacy compromises as well as motivate them making use of privacy preferences available to them.