Personalized Smartphone Wearing Behavior Analysis

Abstract

Next generation smartphones have the ability to sense user contexts such as mobility, device wearing position, location, activity, emotion, health condition. Many apps utilize user contexts to provide innovative services, e.g., pedometer, advanced navigation and location based services. Two of the most important user contexts are mobility patterns (still and walk) and device wearing positions (hand, arm, chest, waist and thigh). We call these two user contexts "wearing behavior". In this paper, we propose a 3-stage framework to recognize smartphone wearing behaviors by utilizing sensor data from smartphones. The framework starts with data preprocessing to extract sensor features and generate ground truths. After the data preprocessing, a threshold based finite state machine utilizes the sensor features to determine whether the smartphone is attached or not. Finally, a decision tree model is built based on the ground truth to determine the wearing behaviors. The experiment results show that our approach can achieve 94 % accuracy in average.