Thesis Defense - Gruteser

Location-based queries are network services that provide users with local information based on their current geographic locations. For example, these services could list nearby restaurants and help users navigate unknown areas. However, these services cause privacy concerns, because they require that users transmit their current positions to an external service provider. Location information collected from GPS or wireless LAN can be so precise that, even if the user issues a query without explicit user identification, an adversary can learn the user's identity by linking the position to publicly available, identified location records. Simply coarsening location information, for example to zip codes, may unnecessarily degrade service quality.

I propose location privacy mechanisms that address this tradeoff between privacy and service quality. For a given population density around a user, automatic resolution control identifies the highest resolution of location information that still preserves a specified degree of anonymity. I present a Linux-based prototype that monitors the density of wireless LAN users and adjusts the resolution of queries so that the query could have originated from other nearby users. The system also changes the network interface identifier to hide a user's movements between different network access points. Simulations and experimental results show that the system preserves privacy while achieving resolutions useful for many services.