Research on Human Motion Recognition Based on Wi-Fi and Inertial Sensor Signal Fusion

Abstract

In recent years, research on human motion recognition has made some progress. However, the commonly-used recognition methods, based on vision and inertial sensors, have certain environmental limitations. In addition, recognition methods based on wireless signals are easily affected by ambient noise. In this paper, we propose a multi-model fusion of Wi-Fi and inertial sensor signals to identify human motion which has fast speed. By wearing an additional inertial sensor on the arm, we can identify seven kinds of basic table tennis movements. Firstly, we constructed the hidden Markov model to classify and identify the Wi-Fi and inertial sensor signals. Then, we used multi-model data fusion technology to fuse and identify the two signals. The fusion method is made stable by randomly selecting the training samples and by changing the number of training samples. After the fusion, the average recognition rate of human motion can reach 97.43%.