The loci semantic location service

Just as coordinate-oriented location-based applications have exploded recently with mapping services, new semantic location services are expected to be critical for the next wave of game-changing applications. There will be a desire for everyday applications to have location-awareness that goes beyond simple numerical latitude and longitude. A new semantic location service layer that abstracts a user's location in the context of places and paths is presented in this dissertation. It automatically learns new places, identifies potentially meaningful places, recognizes registered places, and tracks paths connecting places. Applications can then take advantage of these personally meaningful places and paths registered to the service much as social media applications make use of a user's contact list. This dissertation examines how this semantic location service layer, named Loci, is designed, implemented, and tested. We first study how significant places of an individual can be learned so that a database of personal places can be built. Several place learning techniques are presented that use pervasive radio-frequency (RF) beacon signals, e.g. Wi-Fi and/or cell towers. These techniques interpret locations more in a semantic manner and provide a richer set of contextual information. To keep the service practical, it is augmented by an architecture that controls the active cycles of the sensors based on the user's physical mobility. This results in a location learning and tracking mechanism that is also energy efficient. The entire service layer is completed with a graphical user interface (GUI) to allow the user to correlate algorithm-learned places with semantic significance. The feasibility and usefulness of the entire Loci service layer is confirmed through field tests with real users.