Adaptive On-Device Location Recognition

Location-awareness is useful for mobile and pervasive computing. We present a novel adaptive framework for recognizing personally important locations in cellular networks, implementable on a mobile device and usable, e.g., in a presence service. In comparison, most previous work has used service infrastructure for location recognition and the few adaptive frameworks presented have used coordinate-based data. We construct a conceptual framework for the tasks of learning important locations and predicting the next location. We give algorithms for efficient approximation of the ideal concepts, and evaluate them experimentally with real data.