Activity Recognition Based on Array Sensor

This paper introduces activity recognition based on our proposed array sensor. The array sensor consists of an antenna array on the receiver side and decomposes received signals into eigenvectors and eigenvalues. It exploits these components depending on its applications, such as activity recognition. When an event occurs, the propagation environment changes, and thus the eigenvector and eigenvalue change. Eigenvector and corresponding eigenvalue are inherent to their propagation environments so that to activity as well. Based on the change of these components, we can detect an activity accurately. In addition, using machine learning based on these components, the proposed array sensor can classify several more complex states and activities. This paper also introduces various applications of array sensor.