Motion Detection with Local Linear Embedding and its Application to Indoor Device-Free Human Trajectory Tracking

Device-free indoor human trajectory tracking is critical to support health care applications for elderly people. Many device-free localization algorithms depend on expensive hardware to achieve tracking accuracy. In contrast to such algorithms, this paper proposes a new device-free human trajectory tracking algorithm for indoor environments based on channel state information that is extracted from a Wi-Fi network interface card, which is a low-cost component. The proposed algorithm first uses the characteristics of locally linear embedding to detect whether a person is moving and applies quadratic discriminant analysis to determine the new location of the person. The determined locations of the person are connected to form a trajectory. Experimental results revealed that the proposed algorithm provides an effective solution for passive human trajectory tracking.