Research And Application Of Device Position In Activity Recognition With Smartphone

Activity recognition has become a hot topic with the rapid development of science and technology in recent years, it has attracted more and more attention and is widely used in many fields. With the development of Internet, smartphone becomes popular to humans. Activity recognition with smartphone has become one of the hot topics in the recognition field. Compared to the traditional methods for activity recognition, smartphone is more flexible and convenient. However, there are also many limitations and challenges.This thesis studied the existing methods for activity recognition with smartphone,it mainly focused on the impact of device position upon the activity recognition with smartphone. The research contents of this thesis are listed as follows:(1) This thesis explored the existing methods for activity recognition with smartphone and considered the orientations of device during this process. The differences between raw sensor data caused by the diversity of device orientation were eliminated by coordinate transformation. The position sensitive features were extracted to build a decision tree model for smartphone position recognition.(2) The position-independent features were extracted for activity recognition according to the analysis of the differences between signals of the same activity in different positions. At the same time, the position information of device was used as a parameter to adjust the features which are sensitive to device positions. By feature adjustment,the motion differences between different device positions can be reduced.Position-independent features as well as feature adjustment can make activity recognition with smartphone more effective.(3) Classification model is an important part in activity recognition. In this thesis, we studied several classifiers and explored their performance. Finally, the support vector machine (SVM) model was built for activity recognition which proved to be the most effective model. Considering the energy utilization rate of smartphone,a mechanism of dynamic data transmission was applied in the client to save energy. The client in smartphone can judge the value with the threshold first before data transmission to achieve the goal of saving energy.Experiments confirmed that the scheme proposed in this thesis can solve the problem of the impact of device position in activity recognition with smartphone. It got high accuracies of 93.76% for position recognition and 91.27% for activity recognition. It got good performance in recognition and saved the energy of smartphone at the same time.