MVPTrack: Energy-Efficient Places and Motion States Tracking

Abstract

Contextual information such as a person’s meaningful places (Different from a person’s location (raw coordinates), place is an indoor or outdoor area where a person usually conducts some activity, in other words where it is meaningful to the person, such as home, office rooms, restaurants etc.) could provide intelligence to many smartphone apps. However, acquiring this context attribute is not straightforward and could easily drain the battery. In this paper, we propose M(Move)V(Vehicle)P(Place)Track, a continuous place and motion state tracking framework with a focus on improving the energy efficiency of place entrance detection through two techniques: (1) utilizing the mobility change not only for finding the sleeping opportunities for the high energy sensors, but also for providing hint for place entrance detection, (2) leveraging the place history for fast place entrance detection. We evaluated MVPTrack based on traces collected by five persons over two weeks. The evaluation results showed that MVPTrack used 58 % less energy than previous work and provided a much faster place entrance detection approach.

This work is supported partly by the National Science Foundation under Grant No. 1040725 and No. 0917112.