Ubicomp 2012

R esearchers recently gathered at Tsinghua University in Beijing, China, to share their accomplishments at the 13th annual ACM Conference on Ubiquitous Computing. 1 This multidisciplinary conference featured advances in sensing technologies and activity-recognition algorithms, novel applications, and results from end-user evaluations. Smartphones continued to provide a versatile research platform, facilitating application delivery, sensor data capture, and computational social science on captured data. For many years, capturing sensor data at any scale involved substantial effort and expense. However, dependable smart-phones and robust mobile networks are dramatically reducing data-collection costs. A robust, embedded GPS, for example, lets researchers gather large datasets of human movement on a city-wide scale. A computational analysis of these location traces lets researchers examine previously hard-to-see social patterns in the aggregate. Location Analysis: Tacking Taxis GPS-equipped taxicabs provide continuous , dynamic reports of driver and passenger location as they traverse city road networks. The routes selected and flow of movement effectively function as mobile traffic sensors. Yu Zheng and his colleagues examined how this sensing capability can support urban planning. They learned Beijing traffic patterns using GPS traces from 30,000 taxis and correlated them changes that urban planners had made to the road network. By visualizing the difference between expected and actual changes to traffic patterns, they can help urban planners reflect on the actual outcomes of their plans. In a similar project, Daqing Zhang and his colleagues used GPS traces to learn common taxi trajectories and developed an anomaly-detection algorithm with over 90 percent accuracy. Their algorithm can help identify when changes occur in the road network, which could help commuters respond to dynamic traffic situations and help regulators identify taxi fraud. Looking to optimize taxi-passenger services, Jing Yuan and his colleagues looked to use sensing to make taxis more economic and convenient for both the drivers and passengers. Their algorithm, which compares models of taxi and pedestrian movement, can offer recommendations to drivers as to where to find passengers and can help pedestrians find taxis. As an alternative to GPS, which can be unavailable or unreliable, Richard Becker and his colleagues demonstrated that driving routes can be accurately classified using only cellular handoff patterns. They presented results of their two algorithms, demonstrating that cellular handoff patterns are stable across different routes, speeds, directions, phone models, and weather conditions. Healthcare Applications Researchers at Ubicomp 2012 demonstrated that mobile phones can facilitate the sensing of a great …