LAUSR

The recent proliferation of GPS-enabled mobile phones has allowed people to share their current locations with others. Because disclosing one’s location can be valuable but risky, many services and studies employ a user’s GPS coordinates to determine automatically whether or not those coordinates can be disclosed by comparing them with handcrafted rules or privacy models trained using the user’s actual preferences. However, these approaches that employ GPS coordinates constitute a drain on a phone’s battery when the services assume continuous location sharing. In addition, recent positioning methods (assisted GPS and a WiFi-based positioning) rely on external location providers. That is, when a user’s current location preference is determined using her coordinate point, her location information is disclosed to external providers even if this is not her wish. In this paper, we explore a way of learning a user’s location privacy preference using sensors that are energy saving and that do not rely on external providers. This enables us to save energy and protect a user’s privacy when she is unwilling to disclose her location. Note that the machine learning based approach cannot deal well with a user’s private situations that are not included in its training data. So, this paper proposes a new model that can determine a user’s privacy preferences and handle such outlying situations.