Demo: AD-Sense: activity-driven sensing for mobile devices

This paper designs, implements, and evaluates the AD-Sense system, an activity-driven sensing framework for mobile devices, which aims at providing good accuracy of sensing while reducing the energy consumption. AD-Sense first accurately extracts user's activity, with only acceleration data from user-hold device. Then, according to the activity information, AD-Sense dynamically determines sensor-probing pattern, such as localization, in order that it may use as less energy as possible while meeting the application fidelity. Analysis and testbed experiments show that AD-Sense efficiently tradeoffs the energy consumption and the accuracy requirement, and reduces the hand-operated overhead of user.