Energy-Efficient Motion Related Activity Recognition on Mobile Devices for Pervasive Healthcare

Activity recognition plays an important role for pervasive healthcare such as health monitoring, assisted living and pro-active services. Despite of the continuous and transparent sensing with various built-in sensors in mobile devices, activity recognition on mobile devices for pervasive healthcare is still a challenge due to the constraint of resources, such as battery limitation, computation workload, etc. Keeping in view the demand of energy-efficient activity recognition, we propose a hierarchical method to recognize user activities based on a single tri-axial accelerometer in smart phones for health monitoring. Specifically, the contribution of this paper is two-fold. First, it is demonstrated that the activity recognition based on the low sampling frequency is feasible for the long-term activity monitoring. Second, this paper presents a hierarchical recognition scheme. The proposed algorithm reduces the opportunity of usage of time-consuming frequency-domain features and adjusts the size of sliding window to improve recognition accuracy. Experimental results demonstrate the effectiveness of the proposed algorithm, with more than 85 % recognition accuracy rate for 11 activities and 3.2 h extended battery life for mobile phones. Our energy efficient recognition algorithm extends the battery time for activity recognition on mobile devices and contributes to the health monitoring for pervasive healthcare.