Position-independent activity recognition model for smartphone based on frequency domain algorithm

There are many new issues in human activity recognition using smart phone with built-in acceleration sensors, such as variations of the location and orientation of smart phone. This paper presents a smart phone position-independent activity recognition model based on frequency domain. First, we analyzed FFT curve of Resultant Acceleration in different mobile positions and different activities. The curve shows that FFT results can be used to distinguish different actions. Furthermore, the highest recognition accuracy is achieved under the condition of 39 lower frequency FFT characteristics. In conclusion, recognition accuracy can be improved by 5% while time-consuming reduced by 12.2% in this method.